Trade exploitation

enf91 · 12-30-2017, 05:47 PM

I started the Flyers challenge. After signing some marginally worthwhile free agents, I set up a trade offer to the Bruins.
Flyers send Wayne Muloin, Bob Blackburn, and Claude Laforge.
Bruins send Barry Ashbee.
Assistant: "I think this is a well-balanced offer." (BTW: AI NEVER accepts "well-balanced" offers. Because of this, we have the below.)
I add John Henderson, 34-year-old goalie whose stats are all 1's (actually, I have FoW on, so all I see is, "GOA 1 MEN 1") to the list of Bruins being traded. Suddenly, "It's very likely that they will accept this offer", and I can add the Bruins' 1968 draft picks (both of them) and Derek Sanderson (3.5 star prospect) before it becomes "well-balanced" again.
To check this, I swapped Henderson for some younger 0.5 A/0.5 P players. They didn't have the same effect. I think age still has the same effect that inability did for the AI assessing trade value.

JeffR · 12-31-2017, 08:59 AM

I think I know what's happening there; since you're initially trading 3-for-1, the Bruins' AI is comparing the value of the worst two players to the notional value of the roster spots they'd occupy. If they're better than what it'd expect a replacement-level player to be worth, they add a bit to the trade value the Bruins are getting; if not, they're lowering it. Either way, it's not enough to equal Ashbee. However, once Henderson's added, there's a tangible player for the second-best Flyer to be compared to, and since he's so bad (basically a career minor league backup who's well past his prime) that sends the value the Flyer contributes to the trade up by quite a bit.

I could set a floor on the comparison value for an individual player to the same as the "replacement level" used for an empty spot in the trade. That would close this exploit. The problem is, sometimes a player really does have a value worse than nothing - say, someone permanently injured or with a cap hit wildly out of proportion to their skill level. I'm going to have to think a little about how to avoid that exploit while still accurately reflecting those type of situations.

12-30-2017, 05:47 PM	#1
enf91 Minors (Double A) Join Date: Jun 2014 Posts: 100	Trade exploitation I started the Flyers challenge. After signing some marginally worthwhile free agents, I set up a trade offer to the Bruins. Flyers send Wayne Muloin, Bob Blackburn, and Claude Laforge. Bruins send Barry Ashbee. Assistant: "I think this is a well-balanced offer." (BTW: AI NEVER accepts "well-balanced" offers. Because of this, we have the below.) I add John Henderson, 34-year-old goalie whose stats are all 1's (actually, I have FoW on, so all I see is, "GOA 1 MEN 1") to the list of Bruins being traded. Suddenly, "It's very likely that they will accept this offer", and I can add the Bruins' 1968 draft picks (both of them) and Derek Sanderson (3.5 star prospect) before it becomes "well-balanced" again. To check this, I swapped Henderson for some younger 0.5 A/0.5 P players. They didn't have the same effect. I think age still has the same effect that inability did for the AI assessing trade value.

12-31-2017, 08:59 AM	#2
JeffR FHM Producer Join Date: Apr 2002 Location: Kelowna, BC Posts: 16,623	I think I know what's happening there; since you're initially trading 3-for-1, the Bruins' AI is comparing the value of the worst two players to the notional value of the roster spots they'd occupy. If they're better than what it'd expect a replacement-level player to be worth, they add a bit to the trade value the Bruins are getting; if not, they're lowering it. Either way, it's not enough to equal Ashbee. However, once Henderson's added, there's a tangible player for the second-best Flyer to be compared to, and since he's so bad (basically a career minor league backup who's well past his prime) that sends the value the Flyer contributes to the trade up by quite a bit. I could set a floor on the comparison value for an individual player to the same as the "replacement level" used for an empty spot in the trade. That would close this exploit. The problem is, sometimes a player really does have a value worse than nothing - say, someone permanently injured or with a cap hit wildly out of proportion to their skill level. I'm going to have to think a little about how to avoid that exploit while still accurately reflecting those type of situations.