StatsDetective 4 | EMERGENCY UPDATE for MLB prediction

Our first emergency episode! And not for any reason other than the massive success of my predictive algorithm. In this video, I ramble incessantly and ask more questions than necessary about sports betting.

Let's pick up right where we left off in the last one. I have not significantly changed my methods or strategy. I have not analyzed any new variables. I have made two very minor changes: betting $10 per game instead of $1 and only betting on predicted wins. The first change is cosmetic and makes the good and bad days easier to track. The second change only boosts my win rate from 57% to 60%. So why am I having such improved success?

The reason is multifactorial. First off, even at 57% the strategy was profitable. This was shown in my alpha testing with even lower stakes. Second, the algorithm is most definitely on a hot streak. Across the last 188 baseball games, the win rate has ballooned to an astounding 65%. This level of accuracy is enough to improve modest returns to sizeable ones. There may be an argument that I should return to the old, slightly less successful strategy, to remain under the radar, which brings me to the betting world at large.

In my very shallow internet research into the topic, I found exactly 0 people doing what I am doing. There was no mention of predicting games with any level of accuracy, nobody talking about sports stats that may give an edge and nobody who was even interested in that sort of thing. It is probable that I was simply not in the right corner of that space, that my promised land lay just beyond the horizon, yet I found myself stuck in the hellscape of arbitrage and stale line betting.

I have only a tertiary understanding of these strategies, but from what I gather, they amount to exploiting loopholes in various sportsbooks to turn a profit. I also got the sense that these practices are quite frowned upon and quickly snuffed out by the big bad betting sites. I typically do not align morally with the business practices of large corporations, but in this case, it makes way too much sense. In any case, loophole strategies do not strike me as sustainable practices, but I am certainly not doing that.

The only comment I ran across which hinted at something akin to my methods said that sportsbooks don't particularly mind people who make accurate predictions because they can use that information to squeeze even more money out of all the other bettors. In essence, part of the business model is allowing a small amount of profit to outsource the predictive workload. In that case, I might be able to make modest profits without ever getting banned. But at my current profit level, I think I might be flying too close to the sun, let's talk numbers.

5 days ago, my account stood at $86 and yesterday it was $223. Obviously, that means I had some very good days and very few bad ones. But even on my most average day, with exactly 60% correct predictions, I was up $10. So while a 100% return on investment across 5 days is certainly not sustainable, 10% probably is. In fact, after a rough day yesterday, I was back down to $193 this morning (day 6). But the highs and lows are not my concern, the overall steady gain is, and it looks quite promising.

The question I now face is: assuming my accuracy remains stable over a month or so, do I scale even more? There are several factors, one is that the baseball season does not last forever. Another is that I could be risking a quicker ban if I consistently turn a profit. At this point, I have more questions than answers, but I am hopeful that this experiment has been successful as a statistical proof-of-concept for all the methods I use.