Can this betting scheme survive a random roller??

I need ya'll's advice. I've war-gamed this betting scheme through two different random number generators:

.Net random class and

Cryptographic RNG (MD5 hash- FIPS 140-1) in a craps simulator on the PC.

I started with a buy-in of $2500. The stats I quote below are rigid in the sense that I NEVER played a hunch, nor 'doubled-down' when I got behind, etc. Nothing is fudged in these numbers I report; I don't turn off bets after 2-5 trash numbers, etc., I let ALL the bets stand until I get a resolution.

With the first generator, after 3800 rolls, I'm

*up*$16,800.

With the second (RNG) generator, after 1366 rolls, I'm

*up*$5430. I should explain my betting system and then pose the question.

I base the betting system on the assumptions that the 4/10 are rarer occurrences, and the chance of immediately repeating (bullfrogging) an inside point number before hitting a different point is slim. Accordingly, I don't ever take Odds (or make Place bets) on the 4/10. The bets start out this way:

$10 PL and wait for a point to be established. Let's say the '6'.

No PL Odds (consistent with the 'no bullfrog' idea) $70 -5/9, $72 -8 TOTAL BET: $222

When a Place bet pays (let's call it the '8'), move the $72-Place 8 bet to PL Odds as $70 (skim $2) and (add $6 to the $84 payoff) then divide the payoff $30/30/30 and add $30 each to the 5/9, and 6 so the table now looks like

$100 5/9, $10 PL w/$100 Odds ($70 from the '8' plus $30) but only 3 bets exist now.. On the next hit, regress to $10 PL w/10 Odds and $32 Inside.

If the Come Out establishes the Point as 4/10, bet $220 Inside (50/60/60/50), no PL Odds. On the first Place bet hit, take that Place bet down (either $50 or $60), add the payoff ($70), and divide the total 3 ways. It'll amount to about "$120". Divide $120 three ways, and add $40 to the remaining number 5/9 and $42 to the 6 or 8. So the table now looks like:

$90 5/9, $102 6 or 8 but only 3 Place bets exist now.

On the next hit, regress to $10 PL w/10 Odds and $32 Inside.

*This is the least rewarding scheme*because of course, the Place bets are smaller than if an Inside number was the Point; if I survive to the ISR point and lose there immediately, the profits are at least $135 for hand where the point was 4/10. It means that the recovery rate is about 3-2. (Two $222 PSO losses could be 'almost' recouped with three $135 wins [short $39, but that's if the 3 winning hands consecutively only pay off on the 6/8 twice in a hand--a 'worst case' scenario].)

At multiples of $2500 (my stake), the bets increase by a factor of one... so that when I've got $5,000, if '6' is the point, the bets are $10 PL no Odds, $140 5/9, $120 Place 8; let's say the 9 hits--move the $140 from the nine to PL Odds, divide the $196 payoff 3 ways and spread $60 each, and now the table looks like:

$10 PL w/$200 odds, $200 Place 5, and $180 Place 8.

**, and that's why I take down the 'winning' place bet; in the hopes that I get paid bigger on another number sooner. Sometimes of course, the 6/8 reoccur sooner, but after almost 5,000 rolls the numbers seem to bear out the idea that this ISR takes 3x to pay off 2 PSO's, rather than some of the other 2-1/2 to one ISR schemes,**

*I wanna emphasize that this system postulates that the 4/10 and bullfrogs are rarer occurrences than "other point numbers"***or a guy like me on a bad dice day.**

*AND seems to survive a random roller*With the $2500 buy in, I've survived 7 straight PSO's to recover to $5430 within 1366 rolls (the recovery from $1200 stake took about 15 hands as I recall); the other number generator hammered me as badly, but in two logs, I've not gone bust. It'd take 11 consecutive PSO's to bust out the $2500. Not unknown, but certainly, "rare", and with a 3-2 recovery ratio for an ISR, I wonder if this makes sense.

I'd really like it if someone was willing to wargame this system; if so, please post your results.

My question is, for the mathematicians--does this system hold any validity or (assuming you understand the # generators) do you think the two random number generators I'm using are too predictably patterned and are skewing the results? I wonder if 5,000 rolls is a large-enough sample...

Thank you for your patience!!