Poker Articles

Testing A New System

If you have a new system that you need to try out, there’s a right way and a wrong way to test it. An incorrect testing method can have a disastrous impact on your bankroll. A badly tested betting system, counting method, or strategy table can have long term negative effects ranging from bad habits to bankroll ruination.

With that in mind, I’ve compiled a few tips that should help you test new systems in an accurate way, with minimal risk.

Test Player-VS-House Systems at Micro Limits:
There is no functional difference between nickel limits and hundred dollar limits when you’re talking about Player-VS-House blackjack. So why test at high limits when you can test at low limits for a fraction of the cost? As you can imagine, a system that ends up costing you 35 percent or more of your testing bankroll isn’t unrealistic. Risk the pennies, not the dollars.
Duplicate Realistic Conditions:
Don’t make rules or conditions assumptions that are wildly in the favor of the player. Instead, assume that the actual field implementation of your new system will be under normal, average conditions and seek a test environment that emulates the norm. A lot of casinos are willing to run small stakes tables with much more favorable player odds, in order to get people in the door. Don’t use these for testing, because the results will be skewed!
Test Long Term:
Never field test over the course of a couple of days and accept those results as gospel. You need to test new systems over the course of weeks or months before you collect enough data to make rough conclusions. Relying on short term data is one of the biggest mistakes that people make in the world of gaming, and can lead to not only ruination of the bankroll, but can also lead to dangerous, incorrect assumptions about the way odds work in general.
Record Your Results:
This should go without saying, but people who ‘memorize’ or ‘estimate’ their gaming results constantly amaze me. This horrible practice often turns good data into useless garbage. Record everything! Write down the time of the session, the number of hands played, bankroll before and after, and any interesting notes that you may have.
Compare Your Results to Other Experiments:
When you’ve compiled a large body of data, figure out if the initial results are better or worse than other systems. If they seem wildly out of line, you may need more extensive testing. Either way, if another system yields better results, then you may need to consider adopting aspects of that system, or revising your thinking all together.

It’s only through dedicated experimentation that we learn what is viable under real world conditions. Simulators are all well and good for basic statistical data, but if you can’t realistically perform to the systems specifications, what good is automated data? Happy testing!

By Bill Ricardi
Poker Expert