My friend Wayne told me that it’s important for a writer to be able to “kill his babies.” When he told me this, I knew exactly what he meant. And I was pleased, because when it comes to my babies, I am a brutal serial killer. Take the carcass below for example. It used to have four siblings, quintuplets they were. These clumps of text grew up together under the article title: “Data Minding” which I just sent in to BLUFF Magazine. What these cute little infants had in common was the theme of keeping records at poker. But collectively, they grew too big. There was not enough room for everyone, so someone had to die. Naturally I did what any ruthless son of a bitch would do, and I murdered the weakest one of the lot. They don’t call it the cutting room floor for nothing. It’s bloody down there. If you can stand to look at such a sight, here’s a dead baby:

**When 3 = 370**

So far, I have been using the time unit called “a year” because everyone else does. But really, I have to contort my mind to fit it into that box. Just when exactly does this thing called “a year” start anyway? January first? Says who? Like I’m supposed to wake up on a winter’s day and just because some farmers who lived thousands of miles away and thousands of years ago figured out that celestial motions are patterned and predictable, I’m supposed to feel like today is the start of something? And that the numbers on my poker tally sheet for precisely the previous 365 days hold some special meaning?

Seems to me there’s a better way. If the objective is to use our collected data to “see how we did” over the span of 365 days in order to draw statistically viable, meaningful, and useful conclusions about past performance and future expectations, then yes, I do believe there is a better way. And here it is:

Instead of waiting a year for a year’s worth of data, try this. Tally your results from, say, January 1 to January 1, then tally your results from January 2 to January 2, then January 3 to January 3, and so on. Do that everyday.

Let’s look at what you have after three years. If you do your tallying the traditional way, you’ll end up with at least 3 numbers and at most 5 numbers for each year:

- Your total amount won or lost. (1 number)
- Your total hours played, or your total hands played, or both. (1 or 2 numbers)
- Your rate of winning or losing, which could use the time unit “hour” or “100 hands.” You could do either, or both – for example, if you play multiple tables online, you might want to know your hourly rate, and also your “per 100 hands” rate. (1 or 2 numbers)

So after three years of keeping score, you’d have at least 9 and at most 15 numbers to show for it. That’s not much grist for the number mill. Hardly enough for data mining, more like data dipping. If you want to dig deep into numbers, you need lots of them, like, thousands of them. And doing it my way, that’s what you’ll have:

- 365 = the number of days in a year.
- 365 x 2 = 730 = the number of days in two years.
- 730 = the number of days in a three-year span that have at least 365 days after them. In other words, every three-year span has 730 years inside it.
- 3 = the minimum number of numbers per “year” that you use. (win/loss amount, amount played, win/loss rate.)
- 5 = the maximum number of numbers per “year” that you use. (This applies to those who use both “hours” and “hands” as units for “amount played.”)
- 730 x 3 = 2190 = the minimum # of #’s you’ll have after three years.
- 730 x 5 = 3650 = the maximum # of #’s you’ll have after three years.

Now that’s what I call a deep mine.

## 2 Responses to “When 3 = 370”

So you are basically measuring the rate of change of winrate in what seems a very statistically misleading way. I guess a plot of that value would emphasise just how sick variance is even over the span of several years.

RIP