OOTP + Analytics + AI = Great Success!

[luckymann]

OK, I'm big enough to admit my initial scepticism toward the incorporation of advanced stats was (sorry) completely off base. I'll never be the sort of person who pores over these metrics. Thankfully, with the power of AI, I don't have to.

Please note this is just one man's perspective and still very much a work in progress.

So I spent a long while just collating all of the stats and asked AI to give me layman definitions and what to look for. Then, using this as a guideline, I set up some view tables in OOTP. Sadly, the column situation doesn't really allow you to see these properly (even on a 27" screen, although perhaps sweed's 32" would cater for it), but here's what we finally settled on:

I then got the AI agent to act as if it were the Analytics Director sending a full scouting report to me, the GM.

After a few false starts, this is what it came up with:

Catchers

Frank House is an above-average offensive catcher already, and the profile is stronger than the surface line alone suggests. In 315 PA he has produced 1.8 WAR and a 115 wRC+ with a .366 OBP, .362 wOBA and .463 SLG. The important part is that the underlying quality holds up: his .353 xwOBA is only a shade below the actual mark, while his .524 xSLG and .611 xSLGCON say there is more damage in the bat than his current slugging line fully reflects. He is not just surviving on contact luck either, because the quality-of-contact indicators are strong enough to support it: 86.6 EV, 10.5% barrel rate and 32.4% hard-hit rate, with a healthy 15.3 launch angle and a workable 35.2% fly-ball rate. The approach is sound, with an 8.3% walk rate, 13.0% strikeout rate, 30.5% chase rate and an excellent 82.3% Z-Contact rate. The .336 BABIP is a bit high, but not enough to discount the performance. For us, the takeaway is simple: this is a real offensive asset at catcher, with enough contact skill to keep the floor high and enough impact quality to believe the bat is sustainable.

Infielders

Freddie Freeman is a first baseman giving us well below the level of offensive output the position requires, and there is very little in the data that argues for a near-term rebound. Across 443 PA he has posted -0.9 WAR and a 74 wRC+ with a .305 OBP, .299 wOBA and .370 SLG. The expected data is not much kinder, at .289 xwOBA, .383 xSLG and .478 xSLGCON. That tells us this is not simply a case of poor fortune. The core issue is batted-ball quality and shape. He has just a 5.2% barrel rate, a 24.5% hard-hit rate, 84.8 EV and a very low 7.7 launch angle, while 50.3% of his balls in play are on the ground. That is a bad combination for a bat-first corner profile. His 7.0% walk rate and 18.3% strikeout rate are acceptable in isolation, but not enough to compensate for the lack of impact. The .291 BABIP is normal. The front-office read here is that he is not being sabotaged by luck; he is simply not driving the ball enough to profile as an offensive first baseman.

Jolbert Cabrera looks productive on the surface, but the shape of the profile is unstable and heavily batting-average dependent. In 102 PA he has put up 0.4 WAR and a 111 wRC+ with a .363 OBP, .356 wOBA and .444 SLG, but the expected metrics are far weaker: .254 xwOBA, .380 xSLG and .453 xSLGCON. The biggest warning light is the .410 BABIP. His actual average is .343, but the xBA is only .264 and the xBACON is .314, which suggests he has been getting a lot out of balls in play without the underlying contact quality to sustain it. That contact quality is poor: 81.3 EV, 6.0% barrel rate and only a 15.7% hard-hit rate, with no home runs. He is also not buying himself much margin through plate discipline, walking only 2.0% of the time, though the 15.7% strikeout rate keeps him from being a total empty-at-bat hitter. What I see is a profile that can fake competence in a short run, but one with no secondary skill to keep it afloat once the BABIP comes back.

Enrique Hernández is one of the more stable offensive pieces in the group because the profile is internally consistent and does not depend on any one extreme result. In 387 PA he has 1.8 WAR and a 102 wRC+ with a .365 OBP, .343 wOBA and .404 SLG. His .317 xwOBA is a bit lighter than the surface line, but the .418 xSLG and .504 xSLGCON say the contact he does square up is good enough to support an average bat. He is not bringing much over-the-fence impact, with 3 HR, 86.0 EV, 6.7% barrel rate and 26.2% hard-hit rate, but the approach does a lot of work for him: 11.4% walk rate, 15.0% strikeout rate, 28.1% chase rate and 78.1% Z-Contact. The .327 BABIP is reasonable. This is a discipline-and-competence profile rather than an impact one. For roster construction, he is a useful stabiliser who keeps at-bats alive and does not hurt us, even if there is limited upside in the bat.

Travis Jackson is the highest-upside offensive infielder in the dataset, because the underlying power indicators are already real even though the total offensive line has not fully caught up. In 374 PA he has 1.5 WAR and a 97 wRC+ with a .305 OBP, .335 wOBA and .497 SLG. The power is completely legitimate: 19 HR, .494 xSLG, .599 xSLGCON, 10.3% barrel rate and 27.7% hard-hit rate. His 86.6 EV is solid and the 11.7 launch angle works well enough with a 32.9% fly-ball rate to let the power play. Where the profile is still incomplete is in the overall offensive process. He walks only 5.3% of the time, chases at a 30.3% clip and posts a middling 76.1% Z-Contact rate. That leaves him dependent on damage rather than on getting into consistently good counts or stacking OBP. The .275 BABIP is a little low, so there may be some room for a better overall line, but the bigger developmental lever is approach rather than luck. The takeaway is that the power foundation is there right now; if he sharpens the decision-making a little, there is a materially better hitter in here.

Rick Renick is a more interesting bat than the surface line suggests, but the path to value is narrow and currently blocked by too much swing-and-miss and too many ground balls. In 65 PA he has -0.1 WAR and a 94 wRC+ with a .323 OBP, .330 wOBA and .435 SLG. Under the hood there is some real contact authority: .446 xSLG, .629 xSLGCON, 87.8 EV, 9.1% barrel rate and 27.3% hard-hit rate. That is not replacement-level contact quality. The problem is the rest of the shape. He strikes out 27.7% of the time, walks only 4.6%, chases 38.0% of the time and still puts 54.5% of his balls on the ground with just a 3.8 launch angle. So even when there is impact in the bat, the swing decisions and ball-flight shape suppress the outcome. The .395 BABIP is high, which tells us the current line is not being held back by poor fortune. My read is that there is more raw damage potential here than the current results show, but unless he lifts the ball more and trims the chase, it stays trapped in a fringy game line.

Dave Bancroft is one of the best overall offensive values in the lineup because he combines enough on-base skill with enough playable power for a shortstop, and the total package works even if some of the surface production is ahead of the expected line. In 426 PA he has 3.0 WAR and a 112 wRC+ with a .358 OBP, .358 wOBA and .462 SLG. The expected contact indicators are lighter at .300 xwOBA, .399 xSLG and .469 xSLGCON, so this is not quite as loud as the surface numbers suggest. Even so, the profile has substance. He has 12 HR, a 7.1% barrel rate, 27.1% hard-hit rate, a 10.1% walk rate and just a 13.4% strikeout rate. His chase rate is a good 28.1%, and his 80.8% Z-Contact rate supports the whole thing. The ball is still on the ground too often at 51.7%, which likely caps the power ceiling, but the combination of contact, discipline and enough gap or pull authority keeps him well above water. This is not a star bat, but it is a very playable offensive shortstop profile and one we can trust more than the expected numbers alone might imply.

Billy Hunter is a non-viable offensive profile as currently constructed. In 66 PA he has -0.4 WAR and a 9 wRC+ with a .212 OBP, .199 wOBA and .234 SLG. The expected metrics are better than that, at .261 xwOBA, .365 xSLG and .402 xSLGCON, so he has been poor even relative to a weak baseline. But even the expected line is still below where we need it. He has no home runs, only 77.8 EV, a 1.7% barrel rate and 18.6% hard-hit rate. The one superficially positive thing is the 87.5% Z-Contact rate, but that does not help much when the contact is so soft and the approach is offering almost no on-base support with a 1.5% walk rate. The .220 BABIP is low, so the batting average will likely rise from .203, but this still looks like a bat with virtually no impact and almost no OBP mechanism. The front-office view is that there is not enough offensive substance here to plan around.

Outfielders

Dusty Baker has produced an explosive line, but it comes in only 20 PA, so the right interpretation is signal rather than conclusion. In that tiny sample he has 0.3 WAR and a 174 wRC+ with a .350 OBP, .453 wOBA and .778 SLG. The expected contact is even louder: .488 xwOBA, .826 xSLG and .929 xSLGCON. He has 3 HR already, an 18.8% barrel rate and a 23.2 launch angle with a 50.0% fly-ball rate, so the shape is pure damage. He is also walking 10.0% of the time and striking out only 10.0%. The caveat is obvious: 20 PA tells us very little about the stable level of talent. What I can say is that the quality of contact in this sample is real and the shape is exactly what you would want from a power bat. It is just not yet actionable as a firm evaluation.

Al Smith is one of the true offensive engines in the group, but he is probably outperforming the underlying contact quality to some extent, so we should respect the production without assuming all of it is bankable. In 255 PA he has 2.8 WAR and a 168 wRC+ with a .437 OBP, .444 wOBA and .604 SLG. Those are star-level outcomes. The expected marks are much lower at .363 xwOBA, .473 xSLG and .563 xSLGCON, which still describe a good hitter but not quite this dominant. The shape of the profile explains why he is still highly valuable: 11.0% walk rate, 13.3% strikeout rate, .330 AVG, 13 HR, 8.9% barrel rate and a healthy 46.7% pull rate. He is not a huge raw-impact bat at 86.5 EV and 25.0% hard-hit rate, so some regression in the batting line is likely, but the bat control and approach are strong enough to keep him very productive. My read is that this is a real offensive contributor whose current line may be a bit ahead of true talent, though not wildly so.

Lorenzo Cain is a below-average hitter overall, but one whose weaknesses are specific rather than total. In 325 PA he has 1.5 WAR and an 83 wRC+ with a .320 OBP, .314 wOBA and .386 SLG. The expected line is similar or a touch worse at .276 xwOBA, .360 xSLG and .460 xSLGCON. He has 8 HR, but the underlying impact is light: 80.6 EV, 4.3% barrel rate and 19.2% hard-hit rate. His 20.0% strikeout rate is manageable, but the 7.4% walk rate is only fair and the 73.5% Z-Contact rate is weak, which leaves him with too many empty swings for a player who does not punish mistakes often enough. The .310 BABIP is normal. This is a player whose offensive contribution is not a strength, but the shape is playable if the rest of the value is coming from defence and baserunning.

Leon Culberson is a weak offensive profile and one that looks more likely to regress than improve. In 130 PA he has 0.3 WAR and an 82 wRC+ with a .315 OBP, .311 wOBA and .389 SLG. The expected numbers are much worse: .212 xwOBA, .291 xSLG and .356 xSLGCON. He is hitting .294, but the xBA is only .216 and the xBACON is .264, which means the average is doing a lot of heavy lifting without much quality underneath it. He has just 2 HR, with 79.3 EV, a 2.9% barrel rate and 13.6% hard-hit rate. The approach offers very little buffer, with a 3.1% walk rate and 17.7% strikeout rate. The .347 BABIP is high for this quality of contact. This looks like a contact-luck profile with little authority and little on-base skill, which is not something we should project forward with confidence.

Felipe Alou is close to a league-average bat, but in a slightly frustrating way because there is enough contact quality here to suggest he could be more if the approach were a bit sharper. In 416 PA he has 1.3 WAR and a 98 wRC+ with a .334 OBP, .336 wOBA and .450 SLG. The expected line is a little lower at .297 xwOBA, .439 xSLG and .514 xSLGCON, but still broadly supportive. He has 10 HR, 87.0 EV, a 6.8% barrel rate and a solid 28.0% hard-hit rate with a balanced batted-ball mix. The issue is that the OBP base is thin because he walks only 4.6% of the time, even though his strikeout rate is a manageable 13.9% and his Z-Contact rate is a good 80.3%. He is putting enough good swings on the ball, but not extracting enough free bases to turn that into a more valuable offensive line. I see a stable, serviceable bat with a little hidden upside if the plate discipline improves.

Heavy Johnson is one of the clearest impact bats in the lineup, but he gets there through a high-variance power shape rather than through a complete offensive process. In 403 PA he has 2.1 WAR and a 130 wRC+ with a .370 OBP, .386 wOBA and .558 SLG. The expected metrics tell the same story on the damage side: .444 xSLG and an elite .661 xSLGCON. He has 22 HR, a 12.2% barrel rate and 29.7% hard-hit rate, which is the loudest raw power package on the team outside the tiny Dusty sample. The issue is swing-and-miss and approach. He strikes out 29.8% of the time, chases 32.1% of pitches and has only a 68.9% Z-Contact rate, which is by far the weakest in the group. The saving grace is that his .402 BABIP and .352 xBACON tell us that when he does connect, the ball is hit with enough authority to do real damage. This is a genuine middle-order power bat, but one with more volatility than a cleaner profile like House or Smith.

Front-Office Summary

House gives you a real advantage at catcher.

Jackson is the main developmental upside bat in the infield, while Bancroft is the most stable all-around infield contributor and Hernández is a useful complementary piece. Freeman is a real problem because the position demands more impact than the profile provides, and Cabrera looks like a regression candidate.

In the outfield, Smith and Heavy Johnson are the main offensive drivers, with Alou as a stable average bat and Cain playable only if the non-offensive value is carrying weight. Culberson is the weakest of the outfield group. Baker’s line is electric, but the sample is too small to treat it as anything more than an encouraging early signal.

Starting Pitchers

Bill Holland looks like a back-end innings arm whose surface results are more acceptable than the underlying quality. In 100.0 IP he has been worth 1.3 rWAR with a 101 ERA+, but the estimator stack is not especially kind: 5.37 FIP, 107 FIP-, 4.49 SIERA and 3.91 xERA. That gap tells us he has pitched closer to average in runs allowed than in core strikeout-walk-homer skill. The underlying shape explains why. He is not missing many bats, with a 15.7% K rate and 8.5% K-BB%, and the 27.7% whiff rate is light for a starter without premium command. What has helped him is the contact mix. He is carrying a 1.56 GB/FB ratio, 44.2% ground-ball rate and only 28.4% fly-ball rate, while the 84.5 EV and 12.5 launch angle are both manageable. The issue is that he is still allowing too much damage when hitters elevate, shown by 1.5 HR/9, 16.5% HR/FB and a 7.3% barrel rate. The overall read is that this is a contact-management starter with enough ground balls to survive, but not enough strikeout dominance to trust against stronger lineups unless the homer rate comes down.

Carl Hubbell reads as a sturdier version of Holland: not dominant, but substantially more stable. Over 129.1 IP he has produced 2.2 rWAR and a 102 ERA+, while the estimators are more encouraging than the run prevention line suggests: 5.26 FIP is still high, but 104 FIP-, 4.02 SIERA and especially 3.35 xERA say the underlying contact profile is better than the basic FIP line implies. He is not a bat-misser in the classic sense, with a 17.8% K rate, but his 5.7% walk rate and 12.1% K-BB% are solid. He also keeps the ball on the ground enough, with a 1.42 GB/FB ratio and 45.4% ground-ball rate, while suppressing hard contact reasonably well at 83.6 EV, 7.5% barrel rate and 24.8% hard-hit rate. The red flag is the 1.7 HR/9, which is too high for this shape, particularly with only a 32.0% fly-ball rate. That suggests some home-run variance and possibly sequencing issues rather than a completely broken profile. The takeaway is that Hubbell looks like a playable mid-rotation arm whose true talent is a bit better than the superficial run-prevention line.

Don Sutton is the clear staff anchor. In 161.1 IP he has delivered 6.1 rWAR and a 163 ERA+, and the estimators support that this is not smoke and mirrors: 3.30 FIP, 65 FIP-, 3.83 SIERA and 2.86 xERA. What makes him particularly convincing is that there is no obvious weak point in the profile. He misses enough bats with a 19.3% K rate and 13.5% K-BB%, limits walks at 5.8%, suppresses homers at 0.4 HR/9, and keeps the ball on the ground at a strong 53.4% clip with a 2.04 GB/FB ratio. The contact he allows is also generally manageable, with 84.4 EV, 8.9 launch angle and just a 5.7% barrel rate. The one oddity is a high 26.4% hard-hit rate paired with a low 5.6% HR/FB, but because the rest of the profile is so clean, that does not read as a major concern. This is a genuine No. 1/No. 2 type performance line driven by command, shape and home-run suppression rather than pure overpowering stuff.

Frank Tanana is one of the most interesting arms in the file because the bat-missing and run-prevention indicators are stronger than the workload suggests. In only 71.0 IP he has already been worth 2.5 rWAR with a 157 ERA+, and the estimators are even better: 3.17 FIP, 63 FIP-, 3.48 SIERA and 2.43 xERA. He brings the best strikeout profile of the starters, with a 23.8% K rate, 15.6% K-BB% and a strong 36.3% whiff rate, while keeping HR/9 at 0.4. The trade-off is command inconsistency, reflected in an 8.2% walk rate, and some batted-ball luck, as his BABIP sits at .307 despite only 81.3 EV and 4.0% barrel rate allowed. He is not a pure ground-ball specialist, but 46.7% grounders and a 1.60 GB/FB ratio are strong enough. Tanana reads like a high-quality starter with real bat-missing upside, and if the walk rate tightens even modestly, the profile gets even stronger. This is not just a hot run; the underlying indicators say there is a real front-line starter here.

Relievers

Justin Duchscherer looks like a usable middle-relief arm whose headline line is being flattered slightly by results. In 39.0 IP he has a 127 ERA+ and 0.5 rWAR, but the underlying indicators are closer to average: 4.98 FIP, 99 FIP-, 4.24 SIERA and 4.37 xERA. The strikeout-walk profile is serviceable at 17.8% K, 7.7% BB and 10.1% K-BB%, and he is throwing in moderate leverage with a 1.23 pLi. The best part of the profile is the contact quality allowed. Opponents are averaging only .235 against him, with a .261 BABIP, 84.7 EV and a 10.7% barrel rate that is a bit high but not catastrophic in relief. The concern is the batted-ball shape: just a 1.07 GB/FB ratio and 38.0% fly-ball rate leave him vulnerable if the contact management slips. This looks like a competent bullpen piece, but not one whose 127 ERA+ should be read as dominant talent.

Aaron Fultz profiles as one of the more reliable non-closing relievers. In 38.1 IP he has 1.0 rWAR and a 163 ERA+, and although the estimator line is less dominant, it is still solid: 4.58 FIP, 91 FIP-, 4.08 SIERA and 2.97 xERA. He is not overpowering, with a 19.2% K rate and 30.6% whiff rate, but he controls the zone well enough at 7.9% BB and 11.3% K-BB%. The contact profile is encouraging: .219 AVG allowed, .243 BABIP, 86.1 EV, 7.3% barrel rate and a useful 1.37 GB/FB ratio. The one thing to watch is that 1.2 HR/9 is slightly high for the rest of the line, suggesting a few mistakes have done damage despite generally good contact suppression. Overall, this is a trustworthy left-handed relief option whose performance looks mostly legitimate.
Will Harris is a lower-leverage depth reliever with a fairly ordinary true-talent line. In 45.0 IP he has 0.2 rWAR and a 108 ERA+, but the estimator stack is middling: 4.65 FIP, 92 FIP-, 4.05 SIERA and 3.06 xERA. His 19.6% K rate and 8.2% walk rate combine for an 11.3% K-BB%, which is acceptable, and his 32.3% whiff rate suggests enough bat-missing to survive. He gets a decent number of grounders at 45.3% and a 1.37 GB/FB ratio, and his 83.0 EV allowed is one of the better marks in the bullpen. The problem is that there is no standout carrying trait. The barrel rate is only 5.8%, but the hard-hit rate is still 23.0%, and the overall shape is more stable than dangerous. This is a functional relief arm, but not one who should be pushed up the leverage ladder unless there are no better options.

Al Hrabosky is a high-variance reliever with obvious bat-missing life but too much self-inflicted damage. In just 14.0 IP he has a 155 ERA+ and 0.4 rWAR, but the estimators are poor: 5.90 FIP, 117 FIP-, 4.35 SIERA and 3.80 xERA. The core contradiction is easy to see. He misses bats at a strong rate, with a 20.3% K rate and an excellent 37.8% whiff rate, and his contact authority allowed is not especially scary at 82.0 EV and 25.0% hard-hit rate. But the 11.9% walk rate is a major problem, and the 1.9 HR/9 alongside a 47.5% fly-ball rate make the profile extremely volatile. The 12.5% barrel rate is the highest among the relievers with meaningful sample, which tells us that when hitters do square him up, they can do real damage. He looks like a live-armed bullpen piece with real upside if the strike-throwing improves, but at present he is not trustworthy.

Andrew Miller is the clearest problem arm in the bullpen. Across 35.0 IP he has -0.4 rWAR and a 78 ERA+, with ugly estimators across the board: 6.80 FIP, 135 FIP-, 5.59 SIERA and 6.06 xERA. The entire profile is unstable. He walks 17.3% of hitters, strikes out only 16.1%, and therefore posts a negative -1.2% K-BB%, which is a major red flag and essentially impossible to survive against good competition. Even though the raw stuff still shows through in places — 96.1 FBV, 34.8% whiff rate, 87.1 EV — the command failure overwhelms it. The 33.3% hard-hit rate is also the worst in the file, and the 9.9% barrel rate is high enough to punish mistakes. A 1.62 GB/FB ratio helps a bit, but not nearly enough. This is not a bad-luck profile. This is an arm currently losing the zone and giving away too many free bases to justify leverage.

Patrick Sandoval is another below-average bullpen arm, though less broken than Miller. In 64.2 IP he has -0.5 rWAR and a 73 ERA+, with a 6.25 FIP, 124 FIP-, 4.89 SIERA and 4.82 xERA. He is not missing many bats, with a 15.3% K rate and 5.1% K-BB%, and the 30.3% whiff rate is ordinary rather than dangerous. His 10.2% walk rate is too high for that strikeout level, and his 1.8 HR/9 is a serious issue given only moderate contact suppression at 85.1 EV and 10.1% barrel rate. The ground-ball rate at 41.9% is fine but not enough to erase the home-run risk. This reads as a low-leverage arm whose current results are deserved; there is not much in the line that argues for a better pitcher hidden underneath.

Closer

Huston Street looks like a usable closer, but not an elite one, and his current role value is being carried partly by leverage rather than dominance. In 33.1 IP he has 0.4 rWAR, a 109 ERA+ and the highest leverage in the file at 1.60 pLi. The estimators are mixed: 5.37 FIP, 107 FIP-, but also a better 3.66 SIERA and 3.52 xERA. The strikeout-walk shape is respectable at 19.6% K, 5.1% BB and 14.5% K-BB%, which is one of the better relief command lines in the group. The issue is the long ball. He is allowing 1.9 HR/9 with a 38.8% fly-ball rate and 17.5% HR/FB, which is a dangerous combination in the ninth. On the other hand, the underlying contact quality is not awful: .233 AVG allowed, .240 BABIP, 84.3 EV and 10.7% barrel rate. Street reads less like a lockdown closer and more like a competent leverage reliever whose command keeps him afloat. He can hold the role, but he is not separating himself enough to be untouchable in it.

Front-Office Summary

Sutton and Tanana are the two core rotation assets here. Sutton is the stable ace profile with elite run prevention, strong ground-ball shape and clean estimators, while Tanana has the best pure bat-missing upside among the starters and may still have another gear if the walks settle down.

Hubbell sits in the next tier as a solid, dependable starter whose underlying line is a bit better than his top-line results. Holland is more of a back-end innings piece who survives on shape and ground balls, but whose FIP and home-run profile make him less trustworthy.

In the bullpen, Fultz is the steadiest support arm, Harris is a usable depth reliever, and Duchscherer is playable but probably a touch overvalued by the ERA+ line. Street can close, but the home-run risk means he is more stable-support than dominant stopper.

Hrabosky is the main development bet in relief: there is enough whiff and raw life there to matter, but the walks and barrel damage make him volatile. Miller and Sandoval are the clear liabilities right now, with Miller in particular showing a non-functional strikeout-walk profile.

The decision-oriented takeaway is straightforward: build the staff around Sutton and Tanana, keep Hubbell firmly in the trusted group, use Fultz as the key bullpen stabiliser, and treat Street as a hold rather than an unquestioned closer. Holland is usable but not a priority asset. Hrabosky is worth trying to refine. Miller and Sandoval are the obvious replace-or-rework cases.



I now have this process hard-baked into my ChatGPT account as a Project, so all I need do is attach the CSV of the respective view and use the pre-arranged prompt moving forward to generate these at will.

While I am yet to fully assess the accuracy of this report and will undoubtedly need to refine it further from what you see, hopefully, even at this initial stage, my little MO serves as inspiration for others.

Bravo, OOTP, you've done it again! With a massive assist to ChatGPT.

G

BTW for those interested, here are the source CSV files.

[luckymann]

I envisage using this tool in a number of ways.

One example would be if I was considering acquiring a specific player and wanted to see what the advanced stats were saying about them.

Here's one I had the AI agent produce for Pat Patterson.


Pat Patterson looks like a plausible acquisition target, but more as a complementary piece than as a player to pursue aggressively. In 156 PA he has produced 0.3 WAR with a 95 wRC+, .308 OBP, .332 wOBA and .497 SLG. The immediate read is that the line is unusual: the slugging is strong for a middle infielder, but the overall offensive value is only around league average because the on-base base is thin and the run creation has not separated.

The encouraging part is that the power appears mostly real. His .497 SLG is backed by a .487 xSLG and a .573 xSLGCON, which tells us the damage on contact is legitimate rather than fluky. He has 8 HR in only 156 PA, with an 86.8 EV, 8.1% barrel rate and 30.6% hard-hit rate. For a second baseman, that is a useful offensive shape. He is not an all-fields line-drive type either. The 45.2% pull rate and 25.0% fly-ball rate suggest a hitter whose power comes from turning on pitches and doing damage selectively rather than from broad-based offensive consistency. That can play, especially at the position.

The bigger issue is how narrow the offensive profile is. He is walking only 5.8% of the time, which helps explain the modest .308 OBP, and while the 14.1% strikeout rate is good, it is not elite enough to fully offset the low walk rate. His chase rate at 31.7% is a bit too high for someone without a stronger on-base foundation, though the 80% Z-Contact is solid and supports the idea that he has enough bat-to-ball skill to keep the floor from collapsing. In other words, he is not a bad hitter, but he is a hitter who depends on power doing a fair amount of the work.

There are also a couple of points that cut both ways in projection. His .262 AVG sits below both his .293 xBA and .345 xBACON, which suggests he may have left some batting average on the table. That gives you a mild buy-low signal. At the same time, the .259 BABIP is low enough that some of the mediocre surface line may simply be poor fortune rather than poor skill. That is a positive if you are trying to acquire him before the market fully prices in a rebound. On the other hand, because his OBP is still weak and his value is being driven by slugging, you would want to be careful not to overpay for what is still a somewhat role-specific bat.

The rest of the profile is quietly positive. He has a small positive WPA at 0.54, slightly negative wRAA at -1.0, and positive baserunning marks with 0.3 wSB and 0.8 UBR. That suggests he is not clogging the roster in the secondary areas and can provide a bit of utility beyond the bat. At age 25 with a 55 OVR / 60 POT marker, he also sits in a reasonable acquisition window where there could still be a bit of growth rather than just maintenance.

My view is that Patterson is a sensible target if the ask is moderate and the need is for a second-division regular, strong-side platoon option, or lower-cost infield bat with some pop. I would like him more as a trade target than as a priority free-agent-type cornerstone, because the profile is useful but not foundational. The key argument in favour is that the power looks real and the low BABIP suggests some undervaluation. The key argument against is that the offensive value is still fairly dependent on slugging, with only a .308 OBP and no standout carrying plate-discipline trait.

Front-office summary

Patterson looks like a viable acquisition if the goal is to add a supporting infield bat with real pop rather than to land a lineup anchor. The core selling points are the near-match between .497 SLG and .487 xSLG, the strong .573 xSLGCON, and the combination of 8 HR in 156 PA with a manageable 14.1% strikeout rate. The caution points are the weak .308 OBP, the modest 5.8% walk rate, and the fact that the overall line still only translates to a 95 wRC+. The low .259 BABIP and stronger .293 xBA make him look somewhat undervalued, so there is a buy-low case here. I would classify him as a worthwhile secondary target if the acquisition cost is controlled, but not as a player to push hard for unless your criteria heavily favour middle-infield power.


Again, just a first pass that could do with some refinement, and which is very much limited by the inability to produce these advanced stats at a league for context (devs, thoughts on this being introduced??), but it certainly moves the needle from where I'd be doing it manually.

[luckymann]

The third and, arguably, most useful application for this is what we'll call a "target search", whereby I give the AI agent certain criteria and ask for its best selections that match them.

So, in this example, I gave the AI the full league player list and asked it for the best 10 groundballer LH relievers.

Here's what it came up with:


Here is a target search based on the attached leaguewide pitching file, with the pool filtered to left-handed relievers and then weighted toward true groundball utility rather than just raw GB%.

I did not simply take the 10 highest ground-ball rates. I prioritised pitchers whose groundball profile actually looks usable in a front-office context. That means GB% and GB/FB were the starting point, but I also weighed K%-BB%, HR/9, FIP-/xERA, contact quality allowed, workload and leverage. In practical terms, I was looking for left-handed relievers who keep the ball on the ground and still look like real pitchers, not one-dimensional sinker specialists who are surviving on noise.

For meaningful evaluation, I also leaned away from tiny-sample names and toward pitchers with enough innings to say something useful.

Top 10 Groundball Left-Handed Reliever Targets

1. Tim Hill, STL
Hill is the cleanest fit for the brief because he combines elite groundball shape with genuinely strong run-prevention indicators. In 31.0 IP he owns a 56.5% ground-ball rate and a 2.18 GB/FB ratio, both near the top of this subset, but the key is that the rest of the line supports it. He has a 213 ERA+, 80 FIP-, 2.22 xERA, 0.3 HR/9 and just a 4.5% HR/FB. That is exactly what you want from a groundball lefty: not just grounders, but a real ability to suppress damage. The contact quality is excellent at 81.2 EV, 1.2% barrel rate and 11.8% hard-hit rate, which is one of the best weak-contact profiles in the file. He is not overpowering, with a 19.2% K rate and 6.2% K-BB%, but for this role he does not need to be. The 1.65 pLi also tells us he is already being trusted in leverage. This is the best blend of fit, performance and specialist utility in the pool.

2. Brian Fuentes, CHC
Fuentes is the best version of a groundball reliever who still misses bats. He is not quite as extreme on grounders as Hill, but 47.9% GB and a 1.47 GB/FB ratio still make him a legitimate fit, and the bat-missing component is much stronger. In 49.1 IP he has a 27.9% K rate, 15.9% K-BB%, 40.0% whiff rate, 181 ERA+, 79 FIP- and 2.67 xERA. That is a real late-inning profile. He is also keeping the ball in the yard at 0.7 HR/9 with only a 10.5% HR/FB. The contact quality is good rather than elite, with 83.7 EV, 6.8% barrel rate and 23.9% hard-hit rate, but that is more than good enough paired with the strikeout ability. If the goal is not just groundballs but a left-handed reliever who can handle leverage, Fuentes is one of the best targets in the file.

3. Leo Kiely, PIT
Kiely makes the list because he is a true workhorse groundball lefty with a stable, useful shape. In 101.0 IP he has a 51.8% ground-ball rate, 1.84 GB/FB ratio, 101 ERA+, 86 FIP-, 4.35 xERA and 0.6 HR/9. He is not flashy, but he is durable and the profile is coherent. He walks only 5.5% of hitters, which matters because his strikeout rate is modest at 12.8%. The low strikeout rate keeps him from ranking higher, and the contact allowed is merely acceptable rather than dominant at 84.1 EV, 9.5% barrel rate and 28.1% hard-hit rate. But if the mandate is to find a left-handed groundball reliever who can cover real innings and avoid self-destruction, he is a valid target. He looks more like a steady support arm than a weapon, but those arms have value.

4. Darrell Jackson, SRC
Jackson is a very solid groundball target because the profile is cleaner than the surface name value might suggest. He has 44.0 IP, a 51.1% ground-ball rate and a 1.58 GB/FB ratio, while also carrying a 103 ERA+, 85 FIP- and 3.65 xERA. The shape is good: 0.4 HR/9, 4.4% HR/FB, 83.4 EV and a low 16.5% line-drive rate. He is not a strikeout reliever, with a 15.8% K rate and 5.3% K-BB%, but he is functional enough around the zone to make the contact-management approach work. The barrel rate at 8.6% is a little high relative to the rest of the line, so there is some fragility here, but the total package still reads as a useful left-handed groundball arm. He is not a primary target in the same tier as Hill or Fuentes, but he belongs in the top 10.

5. Randy Jones, BAL
Jones is one of the more role-specific groundballers in the file. The argument for him is obvious: 71.2 IP, 51.5% GB, 1.60 GB/FB, 107 ERA+, 3.25 xERA and only 83.7 EV allowed. He is also not beating himself, with a 5.5% walk rate. The argument against him is equally obvious: he has very little bat-missing ability, with only a 9.4% K rate and 3.9% K-BB%. That means he has to live entirely on shape and contact management. To his credit, he mostly does. The 1.0 HR/9 is acceptable, the 9.5% HR/FB is fine, and the launch angle allowed is just 9.8 degrees. This is a pure contact-management groundball lefty. He is not for every roster, but if you want a sinkerball-style relief arm who keeps games under control and lets the defence work, he is a legitimate target.

6. Les Rohr, ARI
Rohr is a very interesting acquisition candidate because the contact-quality profile is strong and the home-run suppression is excellent. In 27.0 IP he has a 48.2% ground-ball rate, 1.74 GB/FB ratio, 165 ERA+, 81 FIP-, 3.23 xERA and only 0.3 HR/9. The quality of contact allowed is among the best in the pool: 81.4 EV, 4.8% barrel rate and 22.9% hard-hit rate. He is not an especially dynamic bat-misser, with a 16.5% K rate and 6.1% K-BB%, but he has enough of a floor to make the groundball profile useful. The only reason he is not ranked even higher is that the sample is smaller and the line-drive rate at 24.1% is a bit higher than ideal. Still, this is a strong target if you want a lower-cost lefty whose profile may be better than his name value.

7. Dean Stone, SD
Stone makes the list because he offers one of the cleaner weak-contact groundball profiles in the group, even if the sample is still modest. In 23.1 IP he has a 46.4% ground-ball rate, 1.78 GB/FB, 143 ERA+, 85 FIP- and 2.48 xERA. His contact suppression is excellent: 83.4 EV, 1.4% barrel rate, 13.0% hard-hit rate and just 0.4 HR/9. He does not overpower hitters, with a 16.5% K rate and 4.1% K-BB%, so this is not a dominance play. It is a shape-and-contact play. The 1.48 pLi is notable, because it suggests he is already handling meaningful spots. The sample size is the obvious caution, but in terms of pure groundball lefty functionality, he deserves the placement.

8. Bob McClure, LAD
McClure is one of the most groundball-heavy arms in the file, and that gets him onto the list despite some real concerns. In 33.0 IP he has a 57.3% ground-ball rate and a 2.46 GB/FB ratio, both elite. He also has a decent 20.2% K rate and 32.9% whiff rate, so on paper there is more than just sinkerball volume here. The problem is the quality of contact when he does get hit: 86.0 EV, 34.0% hard-hit rate and a 12.5% HR/FB. His actual performance line is weak as well, at 79 ERA+ and 4.18 xERA. So why include him? Because true groundball lefties with this level of batted-ball tilt and enough strikeout ability are not common, and if the contact quality can be stabilised at all, there is something to work with. He is more of a projection or refinement target than a polished acquisition.

9. Braxton Garrett, BAL
Garrett is a very good example of why this search cannot be run on GB% alone. On one hand, he is exactly the sort of left-handed groundball reliever clubs look at first: 49.7% GB, 2.29 GB/FB, 39.1 IP, 84.7 EV and only 4.9% barrel rate. On the other hand, the run-prevention line is mixed: 80 ERA+, 103 FIP-, 4.60 SIERA, 3.88 xERA and 1.4 HR/9. The reason he still makes the top 10 is that the shape is strong enough to believe there is still usable value here. The strikeout and walk line is fine, with a 15.8% K rate, 8.4% walk rate and 7.4% K-BB%, and the fly-ball rate is very low at 21.7%. The issue is that when the ball does get elevated, the damage is real, with a 19.4% HR/FB. This is a target if you believe the homer rate is the outlier rather than the skill.

10. Red Bird, SEA
Red Bird rounds out the list because he is a stable, low-walk left-handed reliever whose contact profile fits the groundball brief well enough, even if he is not as extreme as some of the names above him. In 24.2 IP he has a 47.6% ground-ball rate, 1.56 GB/FB, 1.99 xERA and an excellent 14.2% K-BB%. He walks only 3.8% of hitters, which immediately gives him a strong floor, and the contact allowed is good at 82.2 EV, 4.9% barrel rate and 24.4% hard-hit rate. The actual results are less impressive, with an 84 ERA+ and 1.1 HR/9, which is why he ranks lower. But from a skill perspective, there is enough here to be interested. This is the sort of arm whose value may be hidden by a line that has not fully caught up to the underlying control and contact suppression.

Just Missed

Fred Link missed the top 10 despite a 52.4% ground-ball rate because the damage indicators are shakier than I’d want from a true target: 1.8 HR/9, 10.0% barrel rate and only a 6.7% K-BB%. That is a groundballer, but not one I trust.

Antonio Bastardo missed because, while he is probably one of the better left-handed relievers overall, he is not really a groundball specialist in this context. His 40.7% ground-ball rate is too light for the specific brief, even though the strikeout profile is excellent.

Shawn Estes and Doug Davis both had credible cases, but they sat a tier below the final 10 either because the strikeout-walk profile was too ordinary or because the overall contact-management line was less convincing.

Front-Office Summary

If the priority is a true groundball left-handed reliever, Tim Hill is the best fit in the file. He has the strongest combination of ground-ball shape, contact suppression, home-run control and leverage utility. Brian Fuentes is the best higher-octane option because he gives you real strikeout power without sacrificing the groundball fit. Leo Kiely, Darrell Jackson and Randy Jones are the steadier support-tier targets, with Jones in particular as a pure contact-management specialist. Les Rohr and Dean Stone are appealing lower-profile options whose contact quality lines are better than their market perception might be. McClure and Garrett are more developmental or bet-on-shape targets, while Red Bird is a floor play built on control and weak contact. If I were narrowing this to the strongest acquisition targets overall for the brief, the first calls would be Hill, Fuentes, Rohr and Jackson.


I see this as being super handy to whittle down potential FA adds or when looking for undervalued guys around the traps.

If you have your league set up with feeders etc and the setting applied to track advanced stats at all levels, then I suppose you could also use this for assessing potential draftees. But the league would have to be quite mature for this to work effectively, I'd think.

[luckymann]

Obviously, for the first few seasons, you are going to be vulnerable to small sample sizes. But, after a few years, you could have the AI assess them over their career rather than the current season, if you prefer.

This would be where (as I have requested elsewhere) the ability to export individual player stats by season would be really handy, allowing the AI to look at pro- or regressive trends as the player's career has advanced. Hopefully that'll get due consideration for a patched upgrade. No point having these fancy new stats available only to then greatly limit how we can utilise them, is there?

G

[redranger]

Wow. This is really cool. Appreciate you sharing. I’m gonna have to play with this in my fictional league


Sent from my iPhone using Tapatalk

[gpellis]

Very neat. I've been using it as well to basically write a python script to help me manage my entire minor leagues, tell me who should be cut, promoted, etc, and ensure I have the right depth at each level. Makes things a bit easier for me, as I want to manage my full minors but I always get slightly overwhelmed doing it without some sort of guide.

[Guthrien]

I do similar things myself. It's fun, feels like you have a real analytic staff and the stakes are low - the best way to use AI without guilt! The beauty really is there's enough exportable data in OOTP to play with it however you want: analytics in plain language; assessment of current or draft classes; expanded fictional reports on your league or games beyond what you might be used to with OOTP's text generator.

Good examples!

[Guthrien]

Site dragged and double posted me.

[PSUColonel]

Out of curiosity..how do you go about getting AI to do this?

[millhousebc]

PSU.....not sure if it's what the OP does, but I've just taken screenshots and posted to AI, and get them to do an analysis on the data

[luckymann]

Originally Posted by PSUColonel

Out of curiosity..how do you go about getting AI to do this?

OK, so it depends on the agent really. I personally use ChatGPT a lot for my work, so I have a pro version that allows you to start projects. But I think a normal chat would do the same thing albeit less efficiently. Perhaps the same functionality is available across the board.

Just start a chat / project and converse with the agent. Give it as much background info as you want. Ask for its preferred source. Some will ask for HTML, others csv / spreadsheet, some will even do PDF or other types.

Like scouts, AI agents will have different thoughts on advanced analytics, different stats it weighs more heavily, so I'd ascertain that first as you can see I did in the first post.

Once you have that list, set up views that have all of the stats. Unless you have a massive screen, this will likely be unusable for any purpose other than this, although the document itself that you produce can be read outside OOTP.

Do one of these for hitters, one for pitchers. Hopefully they add a few more for defence and you could do one for that as well down the track.

Again, just keep conversing with the agent and developing a "rapport". Get it to look at the table and suggest other stats if needs be to give it a fuller picture, Once you feel you have that down, get it to run the report.

My prompt in this regard was as I mention above, in the style of a analytics staff member to the GM / head of analytics, something like that. No doubt you'll need to refine it. Once you have, give the report a name and tell the agent to file that away. You can then use that as a prompt, so if you called it a Front-Office Report - Hitters, that should serve as the prompt going forward.

AI isn't as scary as people will have you believe. It's kind of like talking to your IT guy at work but on general knowledge. It's very fun if you use it sensibly, but still also frustrating. You get used to the nonsense you must endure on the road to productivity.

For those who are interested, I have attached a few forms I have built using AI for this express purpose. There are fairly detailed sheets covering hitting, pitching, and defence. As I said, the last of these will be of little use for now in an OOTP context.

There's also a "ready reckoner and a more comprehensive list of analytics with definitions for the "deep-divers". Hope y'all find some use in them and feel free to point out any errata.

G

[low]

LLMs... gross.

Sorry not sorry.

[crazyjoedavola]

the only erratum is not thanking you earlier amigo!

[luckymann]

Originally Posted by crazyjoedavola

the only erratum is not thanking you earlier amigo!

De nada.

[ashantewarrier]

Originally Posted by millhousebc

PSU.....not sure if it's what the OP does, but I've just taken screenshots and posted to AI, and get them to do an analysis on the data

That's what I do as well.

[Guthrien]

OOTP let's you export any stat or sortable list very easily. Or screenshots of course. And it's easily doable in a simple chat screen. I've given it some of the work people have done with run value of particular stats like Stuff vs Movement, etc.

It's actually surprising what ChatGPT/Gemini already know about the game if you ask (sometimes it's a version or 2 behind which leads to the next point). I'm pretty sure this website was fed in the training data. Probably Reddit threads about it. You can also tell them to use this site or the wiki.

[WhiskyTango]

Feed AI a GM profile, a team and org overview, see if it will come up with picks and trades which I carry out in Commish mode.

[iowahawkeyedave]

I've done something similar to this in my fictional save.

I use a program called "OpenCode" and just have it point to the import/export directory of my fictional save. So, then I export the roster file for the entire organization with a view that has all the stats, and just tell the AI to analyze my organization from the perspective of an assistant GM.

Another thing I did was took a screenshot of my baseball park's park effects and dropped it into the import/export folder as well, so the assistant GM will also take into account park effects.

It seems to do a reasonably good job of helping me with analysis, like an assistant GM should.

[Icy]

I'm also having fun with ChatGPT and OOTP. I started asking questions but ended developing a few Python scripts to help in Perfect team as a F2P user.

Also for my next solo offline league once all the major bugs are fixed, I have some plans about using the IA to act as virtual assistant to run my team but roleplaying as if it was a human. Also to setup several press guys to ask me questions etc based on the league development, to also help with the roleplay.

As Luckymann said, don't be intimidarted by the AI, even with no coding knowledge you can ask basic questions to it and it will help you.

Also it can really help with roleplay and imemrsion, in fact you can start asking basic questions like, I'm going to play OOTP26, how can you help me with role playing? or to run my team? then based on the answers you can keep guiding it from the very basic to as complex as you want.

[airr233]

I started this last night with Grok where I gave it definitions of every stat, the value of stats, then asked it what it would need to be able to assist me as a GM and you just keep asking questions as to what the AI will need. Then I feed it box scores of each game and ask for it's opinion if any changes should be done. Finally, after every 10 games, I'll give it a season in review where provide stats for the league, minors, to provide an overview as to what moves should be done next. It's been kind of fun and I'm excited to continue it.