Dlan wrote: ↑Fri Apr 09, 2021 9:51 am
I don’t understand how anyone can say the win/loss data, where a team uses a blocking strategy then loses, doesn’t tell us anything.
It doesn't tell us anything because we can't peak into that proverbial parallel universe where we're in the same games and don't donate and then compare results.
Here's another way to illustrate what I'm talking about. Imagine I was psychic and donated every time I knew my opponents had a 4 pt loner. My win % in games I perfectly donate would STILL be significantly lower than in games I don't donate at all becuz in games I don't donate at all I'm obviously getting relatively better cards, hence why I'm not donating. Clearly we would be wrong to say donating is causing the lower winning %. That can't be the case since every time we donate we save our team 2 pts. The real reason the perfect donater is losing more often in games he donates in is becuz he's getting worse cards on average.
We all know this intuitively. The worst cards one gets, the more likely they are going to donate, and the more likely their team is going to lose. But it's the cards that are the primary cause of the higher losing percentage, not the donating. Donating correlates with losing but doesn't cause it. But that said, if one is donating too much, donating will certainly play an ancillary causal role, but even in that case it is still the cards that is the primary contributor to the higher frequency of losing.
An easy example to drive this point home: in a game with all experts, the team that donates up 9-7 is going to lose more often than the team that doesn't donate up 9-7. Does that mean donating caused more losses? No. It just means the experts who didn't donate up 9-7 had better cards than those who did donate (I.E. they were guarded or had a legit calling hand) and that naturally translates to a higher win% for the non-donaters.
Back to my example of the "perfect physic donater" who only donates when he knows he's blocking a 4 pt loner. Now I'm gonna make up some numbers. Let's say in those games he donates once his team wins 50% of the time. In games he donates twice, his team wins 35% of the time, and in games he donates 3 times his team wins 20% of the time, and in games he donates 4 times his team wins 5% of the time. And lets say in games he doesn't donate at all he wins 67% of the time. Well obviously we know we can't just look at these win percentages and declare Not donating at all is better than donating once, and donating once is better than donating twice, etc. We know that reasoning is fallacious. The only way to really see the positive effect of the psychic donater's strategy would be to peak into that proverbial parallel universe and compare the results of the games he donated in this universe to the exact same games where he doesn't donate in the other universe. Then we would quickly see the value of his psychic abilities whereas before if we just stuck to this universe and generically compared his win% of games he donates in to games he doesn't we end up incorrectly assuming donating is a problem.
Ok, I hope that contrived example drives my point home. Win/loss% is very poor evidence to look at when deciding the efficacy of donating. Evidence so poor and potentially misleading that we are better off ignoring it all together if we wish to keep this analysis serious.
Dlan wrote: ↑Fri Apr 09, 2021 9:51 am
Not all, but many of our OE games are close. Giving an opponent 20% of the points needed to win doesn’t matter? Blame the cards?
Donating is not actually "giving an opponent 20% of the points needed". On average it's a fraction of a point. How much of a fraction depends on the nature of one's donating hand as not all donates are created equal. And when exactly it's worth it for a team to "pay" that fraction is absolutely up for debate and probably debatable forever without a very good computer simulation.
Dlan wrote: ↑Fri Apr 09, 2021 9:51 am
The evidence is clear, we over-use donations.

I think the evidence is strong enough that it certainly has me rethinking a lot of spots. I just need a larger sample size on certain hand configurations to help me further refine my donating strategy. I really need to start doing kitchen table donates to speed things up. Add it to the long list of samples I need to do.
The whole point of my post though is to get us to recognize where the evidence is coming from. It's not the win/loss% that matters, it's the EV of donating that matters and so far after a decent set of hands, donating is not doing well at all. The EV of donating is clearly and significantly negative. I still think there's some runbad cooked into my individual sample, but the evidence is still strong and undeniable enough that I know I have to make some changes to my game.