Why AI teams are reluctant to trade away draft picks

May 16, 2014-

As has been noticed, it can be quite difficult to convince an AI team to part with draft picks, even if they are a "contending" team. It takes a little more than a Tweet to explain, though.

The fundamental problem is that the AI is bad at predicting the future. It can't reliably predict if a team is going to be good or bad in the future, mainly because it can't predict what will happen in trades and in free agency. So for the AI, there is a lot of risk in assuming its team will be good for a while and trading away its draft picks, since it could actually wind up being bad in which case those draft picks will be very valuable (see: the Brooklyn Nets in a couple years).

This problem is compounded by the fact that there is a human GM. The AI knows to be wary of the human's trades, since humans are known to do devious things to outsmart the AI. For instance, making trades to take the AI's draft picks while also making the AI team worse. Or tanking a couple years and trading away seemingly valuable draft picks to gullible AIs, only to have them turn into late picks. In theory, all of these tricks people come up with would not work against a truly smart AI. In practice, the end result is basically the same: AI teams need to value their own future draft picks very highly because the future is uncertain.

There is one exception to AI teams overvaluing draft picks. After most of the season is completed, the AI has a good idea about where its pick will land in the upcoming draft. AI teams are much more willing to trade away picks when they are confident about where those picks will fall in the draft and about how good the upcoming draft class is. Unfortunately for human GMs, this means that really good picks in the upcoming draft will still be hard to trade for because the AI knows they're really valuable. But later picks are much easier to get in the current draft than in future drafts.