For the last few seasons I have, at various times, tried different systems based on teams being on winning/losing streaks and them either continuing or stopping etc. Generally I would give up after a month or so at somewhere around break-even. Now, maybe this is because such a system is only viable when used as a starting point to then study the individual games to decide which, if any, to bet on
- I say balls to that I hate studying
So, I tested a few ideas last year and then paper-tested for the last 2 months of the season and there were some promising results. I packed it all away for the winter, but I have now dusted it down and ready to set it loose on the new season.
The problem with betting on every team on a streak, whatever your definition of a streak, is that you will always bet on that team until the streak stops, ie you know for certain that you are going to betting on a lot of losers. This made me think of my financial trading and the old saying that the safest most reliable way to make money is to catch the middle 50-60% of any trend(a streak is no more than a trend after all). The 1st 20% can be costly because so many trends will fail in this early time-frame that you will be having vastly more losing trades than winners.
The last 20% or so can be dangerous because greed takes over and also because you obviously never actually know when the trend will end. So, people hang on long after they should have sold, hoping that the trend will resume anytime soon and everyday giving back more of their profits, or worse.
So, in baseball terms I wanted to find what was the most common length of winning/losing streaks and try and concentrate on betting on a few games around this point instead of hoping to catch that team that wins 16 in a row.
This eventually led me to come up with the following parameters for betting: -
1. teams on a 3 or 4 game losing streak playing teams on, at best, a 2 game winning streak, or at worst, a 2 game losing streak
2. teams on a 3 or 4 game winning streak playing teams on, at best, a 2 game winning streak, or at worst, a 2 game losing streak
3. teams on a 3,4 or 5 game losing streak playing teams on at least a 3 game winning streak
I used these for the paper tests for the last 2 months of the season and the results were small to moderate profit and yield - ok, but nothing too take forward into next year. Closer analysis found something, that I guess shouldn't have been too surprising - teams at odds against produced far more losers. Far more than they should have done really and as I looked further I could see that there was a cut-off point around 1.85-1.95 - above that there were far, far more losers than winners.
So, now I had a filter for the system - I would only bet on teams whose odds were below 1.9 (the odds I used were the average odds from Betbrain for around 20-25 bookies).
Of course there is only data from around 120 games used for this analysis, which is not enough to be accurate. Plus the last month of the season could be said to be a different beast because of the effects of the larger rosters and teams trying out youngsters. However, I am going to give it a go as it's something to fill the hours of day
I have gone on a bit here, so I will summarise the system in a separate post.
Baseball Streaks XXIII: The Final Fight
.GIF "Thumb(1)")
