The Big Bankroll Building Challenge

[2low]

Anybody ever chart your wagers? I really like creating charts as a visual way of evaluating my work. I use them for poker and trading activities, and thought it may be interesting/useful to look at a chart of my horse playing as well. It is pretty easy for me to get caught up in where I am now, but a chart shows visually how I got here.

This is a chart of the "Phase 1" test of my black box. This test was only 397 wagers, but it is interesting how I rolled around just above break even on average until one big score.

I'd love to think this is what I can expect - break even for a while, hit a big number and break even from there, rinse and repeat. I doubt it will be that easy.

[pondman]

Can you throw in a linear regression line?

[pondman]

I've been utilizing graphs in handicapping since the 80s. From a glance your handicapping needs some work. I'd be more than happy to give you a few suggestions on a method that would have a more upwards possibility-- playing single win bets in New York comes to mind.

[timtam]

Pondman,

How about giving the rest of us a few suggestions ? .
I play the winter meet at Aqueduct and also play single horses. Might be a
nugget there for some of us.

[2low]

Originally Posted by pondman

Can you throw in a linear regression line?

Interesting suggestion. This is excel, and I haven't ever tried it, but I'll try to look into it this weekend if I have time.

[FakeNameChanged]

Originally Posted by 2low

Anybody ever chart your wagers? I really like creating charts as a visual way of evaluating my work. I use them for poker and trading activities, and thought it may be interesting/useful to look at a chart of my horse playing as well. It is pretty easy for me to get caught up in where I am now, but a chart shows visually how I got here.

This is a chart of the "Phase 1" test of my black box. This test was only 397 wagers, but it is interesting how I rolled around just above break even on average until one big score.

I'd love to think this is what I can expect - break even for a while, hit a big number and break even from there, rinse and repeat. I doubt it will be that easy.

2low, How large were your test bets around the 375 mark? You went from roughly -$30,000 to a plus 70,000 in a few or even one bet. That's a huge swing. The thing to consider if you're live, would you have the stomach to continue when you're 40,000 DOWN? With a half mil, or one mil bankroll, it probably wouldn't bother you. I thought you said these were flat bets or pct. of bankroll, but those wild swings look like some form of increasing bet size.

[2low]

Originally Posted by Whosonfirst

2low, How large were your test bets around the 375 mark? You went from roughly -$30,000 to a plus 70,000 in a few or even one bet. That's a huge swing. The thing to consider if you're live, would you have the stomach to continue when you're 40,000 DOWN? With a half mil, or one mil bankroll, it probably wouldn't bother you. I thought you said these were flat bets or pct. of bankroll, but those wild swings look like some form of increasing bet size.

These were all $2 bets. I went from down $30 to up about $70. Most of it was one horse that paid a total of about $95 for a $2 WPS bet.

Stomach won't be a problem if this succeeds and I am able to build. If I am good at anything, it is thinking about money properly with regards to gambling. It isn't money to me while in the bankroll, it is betting units. A semi-pro poker career has taught me plenty about that. You simply can't be good at poker if you think about how much money you have to call instead of how many betting units you have to call relative to the pot and stacks. Being properly bankrolled allows proper thinking and proper action. It can't be different in horse racing.

I've got my new chart set up to track betting units instead of dollars, so I'll be comparing apples to apples if I am able to increase bet size during my 2,000 bet test.

[2low]

Speaking of my 2,000 bet test, as I expected, this is off to a very slow start in terms of wagering activity. When I added my last factor, the first impact was to drive my sample size down to a point where I simply don't have enough races booked to bet at most track/surface/distance combinations. Plus, established tracks are closing up shop for the winter while new winter tracks are coming on line. It takes time to log the races necessary to identify wagering opportunities.

One reason it takes a while is that I toss out the biggest winner in each bucket, and consider it a loser. Then, with the rest I need to see a minimum net profit excluding the tossed winner. My hope is that this will cut down on the number of wagers I make on false positive opportunities.

Had I not tossed the largest winners, I would have probably 4x or 5x the number of bets available. However, if they are real opportunities, they will eventually catch up without the biggest winners.

[FakeNameChanged]

Originally Posted by 2low

These were all $2 bets. I went from down $30 to up about $70. Most of it was one horse that paid a total of about $95 for a $2 WPS bet.

Stomach won't be a problem if this succeeds and I am able to build. If I am good at anything, it is thinking about money properly with regards to gambling. It isn't money to me while in the bankroll, it is betting units. A semi-pro poker career has taught me plenty about that. You simply can't be good at poker if you think about how much money you have to call instead of how many betting units you have to call relative to the pot and stacks. Being properly bankrolled allows proper thinking and proper action. It can't be different in horse racing.

I've got my new chart set up to track betting units instead of dollars, so I'll be comparing apples to apples if I am able to increase bet size during my 2,000 bet test.

Haha, sorry on my phone, I didn't see a decimal . Thought it was 20000. No TUMS needed.

[ultracapper]

This is an interesting thread. Thanks for including the board.

[overthehill]

for what its worth. I think there are a lot of factors that go in to bet sizing.
and i dont think its a one size fits all thing at all. I remember one day years ago I lost 8% of my bankroll and it really bummed me out. I thought i had some really good wagers but when all was said and done losing 8% in one day was way too much for me. So the questions would be how many bets do you make on a given day. what do you perceive the likeliho0d of the horse winning and how much of an overlay you perceive it to be. I am curious how you determine your place and show strategy. There was a guy in the parlor the other day and he loved the way a horse looked on the track but the horse was over 50-1 at the time so he bet it $12 to place and $20 to show. the horse won the race at 70-1! this is a guy who sometimes bets $40 cold triples.

[2low]

Originally Posted by overthehill

for what its worth. I think there are a lot of factors that go in to bet sizing.
and i dont think its a one size fits all thing at all. I remember one day years ago I lost 8% of my bankroll and it really bummed me out. I thought i had some really good wagers but when all was said and done losing 8% in one day was way too much for me. So the questions would be how many bets do you make on a given day. what do you perceive the likeliho0d of the horse winning and how much of an overlay you perceive it to be. I am curious how you determine your place and show strategy. There was a guy in the parlor the other day and he loved the way a horse looked on the track but the horse was over 50-1 at the time so he bet it $12 to place and $20 to show. the horse won the race at 70-1! this is a guy who sometimes bets $40 cold triples.

My strategy is strictly numbers based. My goal is to create a black box that does all the work for me. While I love watching the horses run and while I like looking at the form, what is most intriguing to me is writing a program that can beat the races by itself. I have an interest in math-based gambling games, and I have an interest in coding, even though I have no training and no programming background at all. This marries the two interests perfectly.

My first goal was to write software that actually works, and I did that. It is an on-going effort, but it clearly runs correctly when I push the button. It may be the worst programming language of all time, and a real programmer would vomit if he/she saw my code, but I don't care about any of that.

Goal two is to actually beat the races using my working program. That's what I am testing now, and I'm sure I'll keep tweaking things along the way.

Goal three is to buy an island with a PayPal transfer.

Bet sizing for me at this point is simply betting the minimum until I make money. I am in no rush to make the maximum. If the program works, the money will follow, and bet sizing will increase. At this time, I am betting win, place, and/or show at 1% of total bankroll or $2, whichever is greater. With more data, I may find that I should bet more or less than 1% on any of the three wager types. I am also tracking exacta/trifecta performance and may add those at a later date.

Deciding whether to bet Win, Place or Show is again not my choice. The numbers tell me (well, my software) what to do. If a set of conditions nets at least $12 cumulatively in the win column, I bet win when those conditions arise. For Place, it is $6, and for Show it is $3. This is after the biggest win is assumed to be a loss - just a little more buffer. If those conditions are met, the program spits our a sheet with a wager on it, and I upload it into my ADW account. I have no idea why I am betting, other than the program told me to. I do spot check to make sure my software is following the rules though.

There is no magic to those numbers by the way. I pulled them from my nether regions. But, I think they provide some insurance against bad bets.

[2low]

Originally Posted by ultracapper

This is an interesting thread. Thanks for including the board.

Thanks! Hopefully it will be entertaining and good conversation.

[2low]

Ug.....the hazards of unique data gathering. And using...

It turns out that the final piece of my puzzle is being used wrong. Instead of using the factor (or whatever you'd call it - it is a calculation) for the current race, my program is grabbing the factor from the previous race and using it against the current race. Obviously, this is a problem, but it is also a big problem. So big in fact, that I'll need to re-run all races for all tracks to get it right. I see no sense in wagering until I get this fixed.

Luckily, it is an easy fix, but it will take a while to get it applied to all tracks. At least there are not too terribly many tracks affected that are running now.

It had no impact on the first test by the way.

[rsetup]

Originally Posted by 2low

My strategy is strictly numbers based. My goal is to create a black box that does all the work for me. While I love watching the horses run and while I like looking at the form, what is most intriguing to me is writing a program that can beat the races by itself. I have an interest in math-based gambling games, and I have an interest in coding, even though I have no training and no programming background at all. This marries the two interests perfectly.

You should look into machine learning and training a model to play the way you would (or, more precisely, the way you would if you followed rules you've established, while allowing for intuition). The end result will be your model will become a better handicapper than you. Read up on AlphaGo as inspiration.