How to tell if a new system is going to work or not

[chilidog]

I don't know about y'all, but I'm always testing out new systems. It's become one of my hobbies.

I used to struggle with knowing at what point is a system probably to fail, or one of the few gems that end up being profitable.

Do you let it run for 100 games? 250? 500? How do you know if results you've had are just part of a larger subset and the results aren't truly indicative of what the actual results will be. Let's say you track 20 bets and you go 12-8 and you're profitable. Should you start betting the system at that point? Or what if you go 8-12 for those 20 bets and decide to scrap the system, even if it would've been profitable over 100 bets.

I leveled up my game over the past few months by adding in a lot of stats and building out models using Python and Excel. One of the things I started tracking was my closing line value (CLV). That tells me really fast if the system is going to be profitable or not.

For example, earlier this month I came up with an idea for betting overs. It's going 6-4 out of 10 bets, profiting +1.64 units. At first glance, it looks like a profitable system to keep tracking. However, the CLV is -3.52%. That tells me already the system is dead. It will end up in the negative at some point. Maybe it'll end up being a good Under system instead.

I have another system that is currently at 8-0 with a -1.42% CLV. It'll probably end up being a losing system.

Here's a really good article on how to start tracking your CLV, using the Pinnacle closer and the SBR half-pt calculator to convert spreads into MLs: https://football-handicapping.com/closing-line-value-on-nfl-spreads-and-totals/

 

[BMR Genie]

BMR has a complete betting calculators. You can visit it from here:

Betting Calculators

Check out 8 different online sports betting calculators to increase your gambling odds including Odds Converter, Parlay Calculator, Round Robin, Spread & Money Line Calculators and more from BookMakers Review.

www.bookmakersreview.com

 

[chilidog]

BMR has a complete betting calculators. You can visit it from here:

View attachment 1787

Betting Calculators

Check out 8 different online sports betting calculators to increase your gambling odds including Odds Converter, Parlay Calculator, Round Robin, Spread & Money Line Calculators and more from BookMakers Review.

www.bookmakersreview.com

Nice! I didn't know that. Is it the same half-pt calc as SBR's? Same data loaded into it? I remember when Granchow set that up for SBR and he kept it updated with seasons worth of data to keep it accurate. I don't know if it's still kept updated or not, or even if it needs to be.

 

[Wagerallsports]

Do you let it run for 100 games? 250? 500? How do you know if results you've had are just part of a larger subset and the results aren't truly indicative of what the actual results will be. Let's say you track 20 bets and you go 12-8 and you're profitable. Should you start betting the system at that point? Or what if you go 8-12 for those 20 bets and decide to scrap the system, even if it would've been profitable over 100 bets.

Are you doing this with all sports or only certain ones?

I'd say the amount of games depends on the sport itself correct?

 

[chilidog]

I do it with all sports. Your CLV tells you everything. Once you get to 10 picks, you'll have a solid indication of which way it's headed. By the time you hit 100 picks, you won't have any doubts about betting the plays or not. 100 picks is when I personally pull the trigger and start betting the picks from that model.

 

[Tuesday]

I do it with all sports. Your CLV tells you everything. Once you get to 10 picks, you'll have a solid indication of which way it's headed. By the time you hit 100 picks, you won't have any doubts about betting the plays or not. 100 picks is when I personally pull the trigger and start betting the picks from that model.

Nice, you should start a thread tracking this for everyone here, I've never tracked those numbers - for hockey and baseball I look at all the usual stats and start from there, then usually just go off superstition and how I feeling at moment haha.

 

[Tanko]

Chilidog,
Whenever I'm testing out a new "system", I usually develop and tweak it over an entire season after back-testing it over several years of data. I try to minimize the adjustments to avoid impacting the new results. If I do make any significant adjustments, I'll back-test it a second time over the prior 3 years of data (if I have it).

If it is working out during the current season, after week ~8 or 9 I'll possibly bet a few games but never fully commit to it until I get fresh/unadjusted results.

I have a number of "systems" that I track for years, even if they are only breakeven or slightly EV+. I may go back and tweak them for each season to see if I can improve performance. Many end up as "dead-ends" and I'll drop them completely after a year.

For NFL, CFB this year, I'm working 7 different evaluations (2 new ones this year). It's a pain to capture and track all this data but it helps me to find new potential areas of focus where an edge might be found.

I also use CLV and +/- units as part of performance measurement.

 

[chilidog]

Chilidog,
Whenever I'm testing out a new "system", I usually develop and tweak it over an entire season after back-testing it over several years of data. I try to minimize the adjustments to avoid impacting the new results. If I do make any significant adjustments, I'll back-test it a second time over the prior 3 years of data (if I have it).

If it is working out during the current season, after week ~8 or 9 I'll possibly bet a few games but never fully commit to it until I get fresh/unadjusted results.

I have a number of "systems" that I track for years, even if they are only breakeven or slightly EV+. I may go back and tweak them for each season to see if I can improve performance. Many end up as "dead-ends" and I'll drop them completely after a year.

For NFL, CFB this year, I'm working 7 different evaluations (2 new ones this year). It's a pain to capture and track all this data but it helps me to find new potential areas of focus where an edge might be found.

I also use CLV and +/- units as part of performance measurement.

I say 'systems', but they're really models. I don't bet on trends or anything that I think is already factored into the existing opener.

I'm tracking 7 different models and actively betting 1 of them. It's my longest-running model (i'm on charlotte -6.5 in the nba for today).

It used to take me way too long every day to update all my data into my spreadsheets. Now it takes about 15 minutes. I paid a developer on upwork to write a Python script for me that scrapes all the data I need and output it to Excel. Then I fire up Excel and load that data into my existing spreadsheet where the formulas spit out what to bet on. I could've had the Python script do everything for me, but then I would've had to tell the developer exactly what I was doing and I didn't want to go that route.

 

[chilidog]

Nice, you should start a thread tracking this for everyone here, I've never tracked those numbers - for hockey and baseball I look at all the usual stats and start from there, then usually just go off superstition and how I feeling at moment haha.

With NHL/MLB, assuming you're just betting on the ML, it's super easy to track CLV. No need to convert the spread into the ML. You can have Excel/Google Sheets do everything for you.

 

[Tanko]

Chilidog,
I figured that's what you meant by system that's why I put it in " ". I don't like to use that term because it sounds gimicky and everyone has a "system". Most are "Always bet a team with 3 straight ATS wins" or some similar stuff.

We are very similar in our processes. Scrape the data, and have excel do the heavy lifting with macros. I say it takes a lot of time due to checking the data and confirming accuracy as well as finding new eval methods.

I am really only focused on CFB and NFL and started MLB a year and half ago. Haven't got into NBA or NHL yet but may someday. I think there are some opportunities in soccer but there's no time to develop the data.

 

[chilidog]

Chilidog,
I figured that's what you meant by system that's why I put it in " ". I don't like to use that term because it sounds gimicky and everyone has a "system". Most are "Always bet a team with 3 straight ATS wins" or some similar stuff.

We are very similar in our processes. Scrape the data, and have excel do the heavy lifting with macros. I say it takes a lot of time due to checking the data and confirming accuracy as well as finding new eval methods.

I am really only focused on CFB and NFL and started MLB a year and half ago. Haven't got into NBA or NHL yet but may someday. I think there are some opportunities in soccer but there's no time to develop the data.

I'd love to develop a soccer model, but I've never been successful with it. 2 of my models that I'm tracking now could possibly be modified to work with soccer; I just need to find a good soccer stats site I can scrape.

 

[Oddsmaker]

All systems even out long term to around 52% - 54% if it is a good one so you need to look at a chart and know when to get off it like stocks.

 

[chilidog]

All systems even out long term to around 52% - 54% if it is a good one so you need to look at a chart and know when to get off it like stocks.

My longest-running model has been active for close to 3 years now. Positive CLV. I'll ride it until it starts a downward trend on my performance graph.