Trading thoughts and ideas.

[dejaqwho]

Seems like trading has a long way to go. I been playing around testing trading, proposing a ton of trades to the CPU.

Computer/Computer trades seem good for 3rd line type players, but numerous times they do awful trades, this is more likely to be the case with top players.

Human/Computer trades are hard to make, almost impossible. Here are some examples.

I tried trading Hemsky for a 4th round pick, nothing.

I tried RNH at trade deadline for what was most likely to end up 10-15 pick, computer said NO WAY, not even close.

Defensmen Rayan Whitney for a 5th round pick...no CPU team would accept.

Some suggestions:

I like the fact that computer says right away we don't want to trade their best young players. But they should at least entertain offers, they should be open to an overpayment.

For teams trying to acquire a player on another team that's in the same division, trade AI should be programed to seek an overpayment, just like in real life.

So far teams are only listing 1 position that they need. In EHM teams were much more detailed on what their needs are. Teams should list more detailed needs... ie. 2nd Line 2 way forward, Checking Center, Playmaking center....etc

In EHM your assistant GM would tell you if the trade proposal was fair or an overpayment, how likely the other team was to accept and that was fairly accurate...and FHM attempts to do this, but they are awful at it, very inaccurate.

[mgom27]

Include Prospects and make Trades bigger in NHL. Because you seen this Offseason Bobby Ryan for then one person or a Draft Picks for Draft Picks.

[Airchallenged]

Originally Posted by dejaqwho

Seems like trading has a long way to go. I been playing around testing trading, proposing a ton of trades to the CPU.

Computer/Computer trades seem good for 3rd line type players, but numerous times they do awful trades, this is more likely to be the case with top players.

Human/Computer trades are hard to make, almost impossible. Here are some examples.

I tried trading Hemsky for a 4th round pick, nothing.

I tried RNH at trade deadline for what was most likely to end up 10-15 pick, computer said NO WAY, not even close.

Defensmen Rayan Whitney for a 5th round pick...no CPU team would accept.

Some suggestions:

I like the fact that computer says right away we don't want to trade their best young players. But they should at least entertain offers, they should be open to an overpayment.

For teams trying to acquire a player on another team that's in the same division, trade AI should be programed to seek an overpayment, just like in real life.

So far teams are only listing 1 position that they need. In EHM teams were much more detailed on what their needs are. Teams should list more detailed needs... ie. 2nd Line 2 way forward, Checking Center, Playmaking center....etc

In EHM your assistant GM would tell you if the trade proposal was fair or an overpayment, how likely the other team was to accept and that was fairly accurate...and FHM attempts to do this, but they are awful at it, very inaccurate.

I've been trying to think of a way to implement the value of picks better. My idea was to assign every pick a value. For example pick 210 (the last pick) would have a value of 1 while pick 1 would have an initial value of 210. There would have to be some heavy multipliers for the top picks and some medium ones for the rest of the first round. Some negative modifiers would have to be applied to later years so that you can't cheat the system like in EA to rack up the top five picks for no real loss.

At the same time there would have to be some modifiers on bottom six players (who are generally less than 3 stars) for teams that REALLY need that level of player. Basically some formula has to be worked out so that every pick or player has a value and teams will accept when they are relatively close in value on both sides.

My only loss is how to value players? Do they have a base value for their skill level and that's it? Is that value adjusted based on the other players of their level and position (Off vs. Def)? If so what stats do you take into account? I think the latter is better but then you would have to think of some equations that pump up the value of over excelling players, but not so much that a 4th liner has the same value as a 1st liner.

[dejaqwho]

Originally Posted by Airchallenged

I've been trying to think of a way to implement the value of picks better. My idea was to assign every pick a value. For example pick 210 (the last pick) would have a value of 1 while pick 1 would have an initial value of 210. There would have to be some heavy multipliers for the top picks and some medium ones for the rest of the first round. Some negative modifiers would have to be applied to later years so that you can't cheat the system like in EA to rack up the top five picks for no real loss.

At the same time there would have to be some modifiers on bottom six players (who are generally less than 3 stars) for teams that REALLY need that level of player. Basically some formula has to be worked out so that every pick or player has a value and teams will accept when they are relatively close in value on both sides.

My only loss is how to value players? Do they have a base value for their skill level and that's it? Is that value adjusted based on the other players of their level and position (Off vs. Def)? If so what stats do you take into account? I think the latter is better but then you would have to think of some equations that pump up the value of over excelling players, but not so much that a 4th liner has the same value as a 1st liner.

Thats a good idea, but instead of applaying an initial linear function from 201 to 1, spare using multipliers and just do a exponential function for value of each pick. So for example top 10 picks would be 600-400 range, next 20 would be 400-200 range, 2nd round would be 100-200 and 3rd 50-100 and so on.

This has already been done by others, but I did some research myself into value of draftpicks from 1995 to 2006 (I am still no done yet). If you look at just the first round 60% of the top 10 picks end up being at LEAST on average(solid 3rd liners/top 6 D/backup G with at least 300 played games) while in the 3rd round its less then 10% and 7th round is <2%. Therefore you need an exponential decrease in value from 1st overall. Also NHL teams Scouting dept do this in real life, they assign values to every pick, based on past statistical data.

For FHM this shouldn't be hard to do because their baseball game does this very well. OOTP baseball has 30 round draft. If you tried to trade from 10th round to 30th round, all 20 of those picks for say a 1st round pick, computer wouldn't do it. So they already have this, I dont know why they just couldn't copy the OOTPB code and paste it in FHM and then just tweak it a bit for Hockey, the math should be the same, just different values.

[drewst18]

This doesn't work exactly for hockey since there are 30 teams in NHL compared to 32 in NFL and value changes but:

http://www.draftcountdown.com/features/Value-Chart.php

That draft chart is an actual chart that NFL GM's use for valuing draft picks.

The balance between fun factor and realism can be tough with these games...

Trades are not common, there are a handful of trades in a year with 1-3 big ones. Rarely ever are they trades involving young can't miss talent like a RNH or John Tavares. While you think that nobody in the league should be untouchable, that is just the "fun factor" side of you talking. When realistically a large number of players are untouchable and teams won't entertain an offer even for these players.

Now obviously If Sydney Crosby was offered for Sam Gagner the Oilers would do it the problem being programming. It is much better for a game to have a stingy AI to an AI that gives everyone away. If you open up the AI to accept every "fair" deal you will be able to find exploits quickly. I would prefer a tough AI in which I don't have to make house rules for myself. Even if it means that I will never be able to trade for a top liner unless hes on the trade block.

Trade difficulty setting could help this a lot.

[Airchallenged]

Originally Posted by dejaqwho

Thats a good idea, but instead of applaying an initial linear function from 201 to 1, spare using multipliers and just do a exponential function for value of each pick. So for example top 10 picks would be 600-400 range, next 20 would be 400-200 range, 2nd round would be 100-200 and 3rd 50-100 and so on.

This has already been done by others, but I did some research myself into value of draftpicks from 1995 to 2006 (I am still no done yet). If you look at just the first round 60% of the top 10 picks end up being at LEAST on average(solid 3rd liners/top 6 D/backup G with at least 300 played games) while in the 3rd round its less then 10% and 7th round is <2%. Therefore you need an exponential decrease in value from 1st overall. Also NHL teams Scouting dept do this in real life, they assign values to every pick, based on past statistical data.

For FHM this shouldn't be hard to do because their baseball game does this very well. OOTP baseball has 30 round draft. If you tried to trade from 10th round to 30th round, all 20 of those picks for say a 1st round pick, computer wouldn't do it. So they already have this, I dont know why they just couldn't copy the OOTPB code and paste it in FHM and then just tweak it a bit for Hockey, the math should be the same, just different values.

That sounds much better. The exponential function was pretty much what I was looking at. Being able to trade the 1st and the 7th for the top pick the 2nd and 3rd for the next pick and the 4th-6th for the third overall made me really bored with EA after a while. It became too tempting to avoid it.

The only thing I could see issues with is valuing position or role players. The one-star fighter is worth more than that to a team that really needs him but is still only one star (or in this case maybe less) to a team that already has someone of his skill set.

[mventres]

Originally Posted by Airchallenged

That sounds much better. The exponential function was pretty much what I was looking at. Being able to trade the 1st and the 7th for the top pick the 2nd and 3rd for the next pick and the 4th-6th for the third overall made me really bored with EA after a while. It became too tempting to avoid it.

The only thing I could see issues with is valuing position or role players. The one-star fighter is worth more than that to a team that really needs him but is still only one star (or in this case maybe less) to a team that already has someone of his skill set.

This can possibly be alleviated with "GM Profiles". In each profile would be weightings for the types of players a GM prefers, whether they draft for value or for need, how they like to balance their lines, etc. More or less should easily implementable...

This would mesh well with similar "Coach Profiles", which define playing style and how the coach plays certain types players in different game situations, regular season vs playoff, etc.

So, if a GM and coach don't see eye to eye, then that will show as well by the GM drafting the wrong players or trading for the wrong types of players...

[dejaqwho]

Originally Posted by drewst18

This doesn't work exactly for hockey since there are 30 teams in NHL compared to 32 in NFL and value changes but:

Draft Countdown - Trade Value Chart

That draft chart is an actual chart that NFL GM's use for valuing draft picks.

The balance between fun factor and realism can be tough with these games...

Trades are not common, there are a handful of trades in a year with 1-3 big ones. Rarely ever are they trades involving young can't miss talent like a RNH or John Tavares. While you think that nobody in the league should be untouchable, that is just the "fun factor" side of you talking. When realistically a large number of players are untouchable and teams won't entertain an offer even for these players.

Now obviously If Sydney Crosby was offered for Sam Gagner the Oilers would do it the problem being programming. It is much better for a game to have a stingy AI to an AI that gives everyone away. If you open up the AI to accept every "fair" deal you will be able to find exploits quickly. I would prefer a tough AI in which I don't have to make house rules for myself. Even if it means that I will never be able to trade for a top liner unless hes on the trade block.

Trade difficulty setting could help this a lot.

Ya definitely, I want the trading to be hard, like in real life. I want teams to have untouchables(but a very small chance that a team might move one of those for a big overpayment), have a bias toward teams in same division.

Speaking of house rules, that was the key for me to keep some enjoyment in EHM 2007 lol. Had to limit myself to never trade for a top 20 pick, self imposed cap of 40 mill, and have to let go of 1 top FA every 2nd year.

[AndyL93]

This is pretty good for the value of draft picks. I'd imagine that (with some tweaking) it could be a pretty good basis for trading down and up etc

NHL draft: What does it cost to trade up? - Broad Street Hockey

[Rob316]

One of the worst things I hated about EHM was the fact every other day there was trades and worst of all by the end of the season the teams literally looked like swiss cheese! By free agency teams let valuable players go for no reason and that went with the prospects as well. I've noticed almost the same thing going on with FHM now.

Trading is going to take some time to polish off in this game there's no doubt, but I like the folks who like the "harder to trade mentality" as this mimics real life and lay's out the initial challenge people are playing this game for is it not ? To actually make decisive decisions rather than just exploding teams apart because they feel 10/12 forwards need to be traded! lol

I would love to see the game feature a slider that relaxes the trading for the people who want to butcher they're teams and be happy still, and a maximum setting that mimics the difficulty in real life situations that make it difficult to move players / contracts to teams looking to shed salary / rebuild mode or loading up for a cup run!

[yzerwing]

Originally Posted by Rob316

One of the worst things I hated about EHM was the fact every other day there was trades and worst of all by the end of the season the teams literally looked like swiss cheese! By free agency teams let valuable players go for no reason and that went with the prospects as well. I've noticed almost the same thing going on with FHM now.

Trading is going to take some time to polish off in this game there's no doubt, but I like the folks who like the "harder to trade mentality" as this mimics real life and lay's out the initial challenge people are playing this game for is it not ? To actually make decisive decisions rather than just exploding teams apart because they feel 10/12 forwards need to be traded! lol

I would love to see the game feature a slider that relaxes the trading for the people who want to butcher they're teams and be happy still, and a maximum setting that mimics the difficulty in real life situations that make it difficult to move players / contracts to teams looking to shed salary / rebuild mode or loading up for a cup run!

Butchering a team isn't even a question at this point. Trading isn't difficult, it is basically impossible at this point. Not complaining as I am sure it is being worked on, just pointing out that trading is not where it should be. Player evaluations are well off at this point and adding picks turns a good trade into a an undo-able trade for some reason. Not to mention that rights of good players get you nothing,

Again, not complaining, just countering your point of the trading being only difficult.

[mventres]

Originally Posted by yzerwing

Butchering a team isn't even a question at this point. Trading isn't difficult, it is basically impossible at this point. Not complaining as I am sure it is being worked on, just pointing out that trading is not where it should be. Player evaluations are well off at this point and adding picks turns a good trade into a an undo-able trade for some reason. Not to mention that rights of good players get you nothing,

Again, not complaining, just countering your point of the trading being only difficult.

Trading SHOULD be difficult, and something that takes time to do. Rarely in real-life are trades determined within minutes. They can sometimes takes months to figure out. It would be awesome if the game can simulate this through negotiations with feedback between CPU and GM...even if it means "I like the offer, but I want to see how our roster does in the first X games first - unless you're willing to make an offer I can't refuse". From one aspect it's frustrating, but the instant-gratification trades simply aren't realistic...

A key part of trading/FA signing/etc is the ability to project into the future, or at least classify players as developing, prime, declining. For trading future draft picks this is a bit difficult as well, but a quick heuristic can be to use the last draft position as a starting point, then adjust for lost player production...

[yzerwing]

Originally Posted by mventres

Trading SHOULD be difficult, and something that takes time to do. Rarely in real-life are trades determined within minutes. They can sometimes takes months to figure out. It would be awesome if the game can simulate this through negotiations with feedback between CPU and GM...even if it means "I like the offer, but I want to see how our roster does in the first X games first - unless you're willing to make an offer I can't refuse". From one aspect it's frustrating, but the instant-gratification trades simply aren't realistic...

A key part of trading/FA signing/etc is the ability to project into the future, or at least classify players as developing, prime, declining. For trading future draft picks this is a bit difficult as well, but a quick heuristic can be to use the last draft position as a starting point, then adjust for lost player production...

Not saying it should be easy, but if you are playing the game at all right now, the trading system is impossible to work with. Adding picks and players turns deals from ok to not. You can add a bunch of 1st round picks and decent prospects and not even be able to land second line type players.

Trades that are considered even, always are turned down and then adding things to those trades rarely gets it done.

[mventres]

Originally Posted by yzerwing

Not saying it should be easy, but if you are playing the game at all right now, the trading system is impossible to work with. Adding picks and players turns deals from ok to not. You can add a bunch of 1st round picks and decent prospects and not even be able to land second line type players.

Trades that are considered even, always are turned down and then adding things to those trades rarely gets it done.

Yes, the current system is, at best, craptacular, and needs a major revision. Trading is a MUCH more complicated thing than people admit. The questions moving forward, as I see them, are:

1. What is a "fair" offer?

Fair is a relative thing, to some respect, as it will be heavily based on team need now and into the future, cap implications, etc. I am not sure to what degree this question can be separated from pt 2 below.

Draft pick trading is another example. Is the 4th overall this year worth the first overall next year for the team current having the first overall pick? (i..e, NSH trading 4th overall this year to COL for 1st round + 2nd round next year)? This depends on the player currently available to NSH, the team (COL) projected placement next year, etc.

2. Does the deal, even if it is fair, make sense?

Fair trades are offered all the time, but maybe they make little sense to make. Even if a seemingly minor trade of a 3rd line player for a top prospect (think Morrow for Morrow PITT-DAL) can sometimes make little sense if it disrupts team chemistry or the 3rd liner is important for a cup run/etc.

Similarly, trading Martin St. Louis to say, FLA isn't a smart thing for FLA unless the cost is basically nothing. However, trading him to say PITT should in theory bring a much higher return, but PITT has cap implications and would like to maintain certain level of prospects/picks/current players/cap space/etc. Moreover, trading St. Louis to a rival may not make much sense either.

Does it make cap sense going forward is another major issue...as are alternative options. Maybe STL is offering Perron to PITT in exchange for something that would make more sense for PITT, than giving up a boatload to land St. Louis...see pt 3.

3. Should negotiation be involved?

Are trades to be made immediately or negotiated over time? Can teams decide over a number of offers, over time, which is the most beneficial now and into the future? Again, cap implications,etc.

[CalgaryCannons]

I don't think I'd call the current system "impossible", it's just very restricted. You can only do one-for-one deals and you won't be able to trade for another teams star players (I'd prefer if it was just very, very difficult to get them in a trade, but it's not that unrealistic).

There are three types of trades I've had some success with:
1. Aging star for top prospect. I'm playing a historical game and traded Mike Vernon for a young Curtis Joseph and Joe Nieuwendyk in his prime for a young Scott Niedermayer.

2. Guy that I don't have room for (and can't send down) for a scrub at a different position or age. Because of waiver rules I occasionally end up with a guy on my roster that I can't send down but don't want. In that case I can usually trade them for a young player who isn't as good (usually a half star lower) but whom I can send to my reserve list. The other option is to find a team that is up against their budget and trade for an aging player (usually the same star rating) who isn't worth their full salary anymore.

3. Outright purchases. I'm rarely anywhere near my budget so I have cash available and have found that teams will happily trade good players/picks for a mediocre player and large amounts of cash. I usually avoid this as it feels like a cheat.

Trading absolutely is something that still needs work, but it can be done.