So DEGEN has been up and running for two weeks now. For those of you who don’t know, DEGEN is a PECOTA-style system for picking games. It’s based on similarity scores, rather than simple power ratings. The strength of the system, to the extent it works, is that it should reveal positive expected value parlays, which are normally an anathema to expected value oriented bettors.
So how has DEGEN done through 30 games now? Well, in Week 1, while my picks went a middling 7-8, DEGEN went a stellar 9-6. Now keep in mind, that’s not 9-6 vs. the spread. That’s 9-6 in parlaying the spread with the over/under. At $100 a bet, a 9-6 parlay weekend would have paid you off $1560. That’s obviously a stellar result – a return on investment of more than 100%.
In Week 2 of DEGEN, I didn’t have time to make straight picks. Instead, I just posted my DEGEN picks, and hoped for the best. How’d they do this week? Well, unfortunately DEGEN didn’t do quite so hot this week, going only 6-8. Here’s the thing though: 6-8 on parlays is still a very profitable week. At $100 a bet, that still pays you out $640, for a return on investment of 45%. So our two week cumulative total, at $100/game, betting parlays only, is +$2200.
The reason of course that a 6-8 week (below .500) is profitable, is that a two team parlay pays out $2.6 for every $1 wagered. The result is that even a terrible overall record of 4-10 on the week, if achieved through parlays, is ultimately profitable. It does mean however that variance may play an even bigger part in this, since one game hit or missed makes such a big difference in the overall payouts.
So overall, the DEGEN parlays are 15-14 through two weeks. Bernoulli teaches us that this is highly unlikely, even over just 15 games, to be simply the result of chance. Assuming a parlay has randomly selected one chance in four to hit, then the odds of going 15-14 (or better) over two weeks is almost insanely small. It’s 0.178%, or about 560 to 1 against. What that means is that it’s probably likely that our true win percentage is greater than 25%.
Now because those rat bastards in Vegas take some vig for themselves, parlays don’t actually pay out 3 to 1. As mentioned, they only pay out 2.6 to 1. At this win rate, for a parlay to be breakeven, you need to win at least 1 in 3.6, rather than 1 in 4. So then the next logical Bernoulli test is that – what are the odds this system is beating the vig. The situation changes, but not as much as you might expect. There is almost still almost a 99.4% chance that this system has outperformed random variance + vig.
So what does 15-14 mean? Honestly, probably not all that much. Just because the system has beaten even regressed variance/vig doesn’t mean it’s going to continue to do so. But it’s at least a good sign.
I’ll try and get a post up soon with Week 6 picks, so we can continue looking at some other ways to take advantage of what, if anything, DEGEN is revealing.