theshirin

We primarily generate the data using Monker Solver, using our own custom trees.

Hi there, thanks for the feedback. Totally agree with you on intro/demo videos, and we're planning to create those in the coming days/weeks.

The features behind a pay wall are the “reaction ranges” for all stack sizes other than 15bb. This is the ability to put in table action and see the resulting data/strategies. Generating that data is very time consuming, requires significant computing hardware, and we continually generate more of it in order to improve the quality of the existing data sets as well as to create more of it (eg more stack sizes). The primary value of our initial offering is rooted in this data, and poker players are willing to pay for it. Historically it’s been prohibitively expensive, though, and a big goal for us was to lower the barrier and make it more accessible.

Our actual moat, though, is the technology and design. The ease of entering and changing table action, how quickly we load the data visualized in all these different ways, the combined aesthetics + logic behind the color schemes… these are the kinds of things we’ve come to expect in consumer technology but hadn’t yet reached the realm of poker tech. We believe that these aspects of Floptimal are groundbreaking within poker.

One of my fav comments of the day, tbh. <3

Thanks for this feedback! We initially thought that experts would be our primary/target audience, but given the reception we're getting from novices, there's clearly value in providing better onboarding for them so that the initial view isn't so overwhelming (although luckily we've seen that even newbies can pick up the charts quickly with a little bit of explanation).

Just to share a little background: our initial designs were of a round poker table but ultimately ditched it for a couple key reasons. We ran into usability issues (eg, the re-raising scenarios were difficult to display clearly, and it took too long for users to read/absorb the summary of action), and the action input wasn't as streamlined (namely, it required far more mouse and eye movement darting around the circle).

It took many iterations to land on what we currently have — the vertical list from early position to late position that eliminates positions in the re-raise rounds — and it proved to be the most efficient for inputting, absorbing, and clearing table action. Ultimately we made this tradeoff to value power usage over first time usage, but we hope to continue improving it and making it easier for first time users as well.

The data was generated in a Nash equilibrium with the sim playing against itself. So the opponent ranges are the ranges you see in the app.