Wouldn't it be better to do a frame-by-frame analysis of an actual TV airing since Price is aired in 60 FPS? CBS.com downscales it to 30, but if you use a good capture device to capture it live on CBS, you should be able to play it back in the better framerate.
In theory, but I don't have that at the moment.
I'm a cable-cutter and my CBS streaming service is only 30fps.
$250 in half a second, so $31.25 every 1/16 (0.0625) seconds. Of course, the money clock is only integers, so the clock likely decreases in a cycle of $31-$31-$31-$32.
The amount subtracted always seems to be in this order: $31-$31-$32-$31-$31-$32-$30-$32, repeated. It's nearly identical to "what it it eliminated $31.25 every time, and then rounded the amount to the nearest dollar", except the last two numbers in the eight-number sequence are respectively $1 higher and $1 lower than they ought to be. You can
almost break it into a perfect quarter-second, 4-number $125 loop, but it fails to correctly predict the last two numbers in every other loop.
That is interesting. I wonder whether the computer running the game is tracking button presses down to the millisecond. So while the monitor can only display one dollar amount per frame during the countdown, when the contestant presses the button, the computer determines exactly what the money should be and displays that number. Then, when the countdown is resumed, the display starts subtracting $31/$32 from the most recent stopping point.
I thought that too, but it doesn't seem to be the case. The same amounts are displayed on every $250 loop, whether the button is hit or not.
So I went to tpirstats to see some of the odd amounts that have been won. I think that revealed the answer.
There are only three sequences, all with perfect $250 loops.
a) One loops perfectly on every $1000/$750/$500/$250 amount.
b) One loops $8 higher, on every $1008/$758/$508/$258 amount.
c) One loops $8 lower, on every $992/$742/$492/$242 amount.
On loop a, the first amount removes $31, going from $20,000 to $19,969, immediately beginning the 31-31-32-31-31-32-30-32 loop. When the clock runs out, it follows the normal $250-$219-$188-$156-$125-$94-$64-$32-$0 pattern.
On loop b, the first amount removes $54 ONLY when starting the clock, going from $20,000 to $19,946, and from then on looping perfectly through that $250 8-number sequence until ending with $258-$226-$196-$164-$133-$102-$70-$39-$8-$0.
On loop c, the first amount removes $39 on the first amount ONLY, going from $20,000 to $19,961, and from then on looping perfectly through that $250 8-number sequence until ending with $274-$242-$211-$180-$148-$117-$86-$54-$24-$0.
Pulled in those playings and counted the money clock frame-by-frame and every playing seems to match up perfectly with one of those three patterns.
And as one final note, all of this applies to all the playings I analyzed from S47 and S48, but at some point it was changed to this pattern from something different. Some S46 playings have winning amounts that appear to be off by about $2, and when I looked at one from 2014 the clock ran much more smoothly, to the point where the tens digit changed every frame and I couldn't make out the ones digit on a freeze frame.