I've been using the Blackjack Apprenticeship training app, and I've noticed that they round in two different ways that compound error in true count calculation: decks remaining shows as 'n' until it drops below 'n - 1' (e.g. 3.1 shows as 4 -- not only does it show as 4, but 4 is what gets used in the subsequent TC calc), and the true count is always rounded down to the whole number. Both of these effects work in the same direction -- they artificially lower the true count. I had a running count of 7, actual decks remaining of just over 3 and therefore reported as 4, which spit out a true count of 1! (1.75 rounded down / truncated to the whole number.) The actual TC in this case is above 2. By their own deviations chart (total of 26), a few of those are triggered at 2. If the same rounding inaccuracies yielded TC of 4 when it was actually 5, no fewer than 5 deviations would be missed. That doesn't seem acceptable when those 5 are of the most important 26 deviations out of more than 100.

My question is this: does anyone know whether their deviation indices take any of these rounding inaccuracies into account in such a way that would minimize their bad effect? Or do they just round the simulation trigger index to the nearest whole number? E.g. if they are recommending or assuming that their followers will use a 'whole number & round down (rather than to the nearest)' approach, which tends to underestimate the true count, dramatically so as DR gets small, do they get some of this back by also rounding the simulation trigger index? E.g. if the simulation says you should start splitting your 10s against dealer 5 at TC of 5.9, and knowing that people might have an actual TC close to 5.9 when their crude estimation is coming out closer to 5, might they report this on the chart as index 5 rather than 6?

I'm asking because I like their chart and the idea of a 'Top 26' deviations, at least to start, but I don't want to use a chart which has 'corrections' to a crude rounding system that I won't be using. Then the corrections themselves introduce large errors for people using a more precise TC calc. I will be using decks remaining increments of closer to 0.25, especially when DR gets low, and rounding TC by no more than 0.5.