Dave's System

[KrapsNovice]

I played the Dave but ran into a series of losses. Thing is once you're in the hole it may take hours to lose it all or get it back. I recall reading on this forum that Dave had spent a dozen hours at the casino. I spent 6 or more hours.

If you do not have the time or enough money to put into this game I suggest you pass. Just some friendly advice does not cost you a thing.

This past month I played with a guy at the Indian casino in Southwest MI, he was playing Dave's method. I asked if he was on this forum because he was playing the Dave. He told me that he did not know this forum. I asked if he came up with this method he did not respond. I laughed he stopped talking normally I never root for someone to lose but I got a little satisfaction when he lost. The table turned hot I switched to the PL he lost quite a lot but I did pretty good.

[nycgags]

Yeah, so I found QUITE A FEW bugs in my simulation, feel proud that I followed one of the only losses on a 100 simulation run down 50 rolls and was able to debug the simulation down to the penny so that my hand written results (using Excel) matched the results of my SQL Query.

After doing so, my first 100 simulation run showed an 18% loss rate which when you do the basic math:

18x(1000) = -18000
82x(+200) = +16400
=================
EV = -1600 (total)
per session = -16

So for shits n giggles I am running simulation 1000 times, it takes about 5 minutes to generate the 100,000 random rolls per 100 sets (10x) = 50 minutes and then maybe another 2 minutes to run the simulation per set (10x) = 20 minutes, so I'll have a better picture in about an hour and a half.

2nd out of 10 returned and it is showing 19 losing sessions, so looks like 17-20% realistic loss rate which is not sustainable (with only a 1k bankroll). Keep in mind, this is also a modified version of Dave's system.

3rd out of 10 returned and it is showing 21 losing sessions (wow) keeps getting worse!

When I finish compiling results from 1k runs I will up balance to 3k and run it to see if the results differ. I assume the loss rate will go down to about 6-8% however the EV will be the same due to 3x the risk (so 3k x 6 still = 18k loss).

[KrapsNovice]

I look forward to your results with the 3k. That is If you do not modify how Dave played it! Are you testing this on the Wizards site or elsewhere.

[nycgags]

I wrote a SQL script (database) to do my simulations.

Results aren't bad actually at 1k bankroll. When you factor in when you bust you still have money on the table so you are not down exactly -1000 but more like -810 or something. The other positive caveat is you never win by exactly +200 but more like +333 so when you factor in these overages, only 1 out of 10 trials of 100 sessions had a negative result:

1. 18/82 (no data, deleted before full 9-run)
2. 19/81 4885
3. 21/79 2695
4. 20/80 3672
5. 19/81 5512
6. 20/80 3295
7. 25/75 (2260)
8. 22/78 1987
9. 18/82 5555
10. 14/86 10552

Average EV per session (average above / 100) = +39.88

Below results all use same roll data set:

1k bankroll run: 21/79; Most rolls = 188 rolls to profit +285
2k bankroll run: 9/91; Most rolls = 875 rolls to bust out with +137 remaining.

Here's the problem going beyond 2k bankroll, you need more than 1,000 rolls to complete simulation to end (either bust or +200 profit)
3k bankroll run: 3/97; 1 of the sessions required more than 1,000 rolls, with ending balance of +1492

I am doing a 4k run but it is taking a very long time, I expect multiple 1,000 roll simulations are occurring.

[nycgags]

re: 4k bankroll, I stopped the sim after running for 36 minutes to see what was going on. Not sure why so slow, 33 runs completed, 1 broke bankroll with +120 remaining it took 329 rolls to bust the 4k bankroll. Most rolls in a session otherwise was 661 for a profit of 232.50. Extrapolated to 100 runs = ~3 busted sessions out of 100 for a ~12k loss. 97 sessions for a 19.4k profit, remember though it takes quite a long time to roll 661 times!

[nycgags]

last post for the night, here is a cumulative 100 session example with a $1500 starting bankroll:

Session - Session PnL - Lifetime PnL

1 $270 $270
2 $233 $503
3 $205 $708
4 $215 $923
5 $220 $1,143
6 $293 $1,435
7 $215 $1,650
8 $275 $1,925
9 $323 $2,248
10 $205 $2,453
11 $205 $2,658
12 $328 $2,985
13 $320 $3,305
14 $225 $3,530
15 $210 $3,740
16 $(1,265) $2,475
17 $235 $2,710
18 $208 $2,918
19 $258 $3,175
20 $310 $3,485
21 $220 $3,705
22 $275 $3,980
23 $325 $4,305
24 $(1,210) $3,095
25 $268 $3,363
26 $275 $3,638
27 $293 $3,930
28 $305 $4,235
29 $250 $4,485
30 $395 $4,880
31 $263 $5,143
32 $250 $5,393
33 $215 $5,608
34 $(1,338) $4,270
35 $248 $4,518
36 $215 $4,733
37 $210 $4,943
38 $270 $5,213
39 $300 $5,513
40 $250 $5,763
41 $290 $6,053
42 $293 $6,345
43 $210 $6,555
44 $260 $6,815
45 $280 $7,095
46 $380 $7,475
47 $250 $7,725
48 $285 $8,010
49 $240 $8,250
50 $228 $8,478
51 $233 $8,710
52 $243 $8,953
53 $215 $9,168
54 $295 $9,463
55 $210 $9,673
56 $(1,343) $8,330
57 $(1,298) $7,033
58 $(1,373) $5,660
59 $203 $5,863
60 $245 $6,108
61 $333 $6,440
62 $293 $6,733
63 $210 $6,943
64 $273 $7,215
65 $233 $7,448
66 $240 $7,688
67 $210 $7,898
68 $245 $8,143
69 $260 $8,403
70 $285 $8,688
71 $223 $8,910
72 $225 $9,135
73 $325 $9,460
74 $(1,368) $8,093
75 $218 $8,310
76 $310 $8,620
77 $243 $8,863
78 $280 $9,143
79 $208 $9,350
80 $(1,383) $7,968
81 $(1,380) $6,588
82 $(1,390) $5,198
83 $243 $5,440
84 $280 $5,720
85 $320 $6,040
86 $220 $6,260
87 $300 $6,560
88 $213 $6,773
89 $240 $7,013
90 $(1,358) $5,655
91 $338 $5,993
92 $(1,348) $4,645
93 $225 $4,870
94 $230 $5,100
95 $220 $5,320
96 $250 $5,570
97 $285 $5,855
98 $280 $6,135
99 $320 $6,455
100 $318 $6,773

[nycgags]

So I had this idea, start with 1500 balance, run the system cumulatively over 100 sessions, any session that drops 1k from start balance - exit session.

This resulted in 20 exits, but overall it earned $3318 here is the PnL after every 10 sessions with (number of exits due to 1k losses since last milestone):

10. $1,530 (1)
20. $2,940 (1)
30. $3,105 (2)
40. $3,458 (2)
50. $573 (5)
60. $950 (2)
70. $1,305 (2)
80. $3,968 (0)
90. $2,983 (3)
100. $3,318 (2)

I am going to make 1 more change, at any point at the start of a session where the balance is < 1500, rebuy for the difference and note the difference in results.

FWIW - Max number of rolls for any session out of the 100 above was 173 rolls, definitely better than some of the 661 rolls we saw before.

With above change on 2nd run, it triggered 6 times at a total cost of $(3215), after 100 sessions 23 1k loss triggered, final balance = 5525 or +810 when you factor in the rebuys and the initial 1500 bankroll. This run had a particularly bad start, first 2 sessions were losing sessions with 4 total in the first 10 sessions.

I will run it 1 more time then I am going back to bed.

3rd run, even more failures (1k losses), 25 in total, after 100 sessions however system still treading water (barely) up +$218.

Seems like more tweaking is in order to make this more profitable - maybe I will play around with it this weekend.

[Moe Bettor]

This can't possibly work over the long run. It is a fixed system that doesn't take into account the fact that there is a bell curve to variation. There are those two ends of the bell where hot streaks and cold streaks reside and the huge middle where you just have chop..which is most of the time at a craps table. If I was a programmer which I am not I'd like set down a chop method of play run over ten thousand times. I play the chop and there are a lot of "ifs". If I see this..I do this..If this happens, I respond this way. May be just too complex to turn into computer code, but what about just following the appearance of inside numbers and points made.

[KrapsNovice]

last post for the night, here is a cumulative 100 session example with a $1500 starting bankroll:

Session - Session PnL - Lifetime PnL

1 $270 $270
2 $233 $503
3 $205 $708
4 $215 $923
5 $220 $1,143
6 $293 $1,435
7 $215 $1,650
8 $275 $1,925
9 $323 $2,248
10 $205 $2,453
11 $205 $2,658
12 $328 $2,985
13 $320 $3,305
14 $225 $3,530
15 $210 $3,740
16 $(1,265) $2,475
17 $235 $2,710
18 $208 $2,918
19 $258 $3,175
20 $310 $3,485
21 $220 $3,705
22 $275 $3,980
23 $325 $4,305
24 $(1,210) $3,095
25 $268 $3,363
26 $275 $3,638
27 $293 $3,930
28 $305 $4,235
29 $250 $4,485
30 $395 $4,880
31 $263 $5,143
32 $250 $5,393
33 $215 $5,608
34 $(1,338) $4,270
35 $248 $4,518
36 $215 $4,733
37 $210 $4,943
38 $270 $5,213
39 $300 $5,513
40 $250 $5,763
41 $290 $6,053
42 $293 $6,345
43 $210 $6,555
44 $260 $6,815
45 $280 $7,095
46 $380 $7,475
47 $250 $7,725
48 $285 $8,010
49 $240 $8,250
50 $228 $8,478
51 $233 $8,710
52 $243 $8,953
53 $215 $9,168
54 $295 $9,463
55 $210 $9,673
56 $(1,343) $8,330
57 $(1,298) $7,033
58 $(1,373) $5,660
59 $203 $5,863
60 $245 $6,108
61 $333 $6,440
62 $293 $6,733
63 $210 $6,943
64 $273 $7,215
65 $233 $7,448
66 $240 $7,688
67 $210 $7,898
68 $245 $8,143
69 $260 $8,403
70 $285 $8,688
71 $223 $8,910
72 $225 $9,135
73 $325 $9,460
74 $(1,368) $8,093
75 $218 $8,310
76 $310 $8,620
77 $243 $8,863
78 $280 $9,143
79 $208 $9,350
80 $(1,383) $7,968
81 $(1,380) $6,588
82 $(1,390) $5,198
83 $243 $5,440
84 $280 $5,720
85 $320 $6,040
86 $220 $6,260
87 $300 $6,560
88 $213 $6,773
89 $240 $7,013
90 $(1,358) $5,655
91 $338 $5,993
92 $(1,348) $4,645
93 $225 $4,870
94 $230 $5,100
95 $220 $5,320
96 $250 $5,570
97 $285 $5,855
98 $280 $6,135
99 $320 $6,455
100 $318 $6,773

NYC,

Dave increased his lays once he lost 1k. His lays were 60/75 and he would increase to 150/175 until he lost 3k. I also increased my lays but I played lower.

Did you increase your lays at any point in the above simulation. Also how much are you wagering on your flats and lays. I enjoy reading your posts.

KN

[House of Orange]

Has anyone thought of adding DI to the mix???

[heavy]

Easy to do in WinCraps. You can enter your own roll files OR just enter a skew in whichever direction you want it to go. Not sure about nycgags programming, though. I'm waiting for someone to figure out what he's missing in there, because the numbers we're seeing here look too good to be true. Something thnick and I agree on!

[heavy]

Yes, boys and girls, it's the revival of the well discussed Dave's System. You'll simply have to go back and read the first few pages of this thread to get a handle on this strategy. Many members of the group have back tested the strategy and shown that it performs miserably over a million or so rolls using various books of rolls available in an assortment of testing programs. I've played the Heavy Dumbed Down (and bankrolled down) version of this many times and done both well and badly with it. Typically when I play it I go with $10 Flat bets and start pushing out $30 lay odds. If I get knocked off one I bump my odds up on the next couple of bets to $60. If I get knocked off a couple more I tend to take all of my odds down and just wait out the roll. Of course, when you take all of those odds down the seven typically shows and you end up kicking yourself for not hanging in there. But if you hang in there the opposite happens. Yeah, it's random craps in action. It is what it is. But check out the thread and give us your thoughts if you're a dark side fan.

[House of Orange]

Often I have wondered, don't know if it's been asked, but if you lay 4&10 instead of 6&8, what would happen??

[Golfer]

I think every darksider should review Dave's play as well as that of Professor H. Heavy, do you still have any Prof H's threads?

Good stuff. If you don't see it early, Dave was barred from a Joliet casino because he was winning too much.

Golfer

[House of Orange]

Shot an email to Steen, see how much he would charge to program the system. Is he still alive? For those of us who can throw a seven it would be an interesting question of profit.

[220Inside]

Shot an email to Steen, see how much he would charge to program the system. Is he still alive? For those of us who can throw a seven it would be an interesting question of profit.

Program it in what? BT? WinCraps?

[House of Orange]

Steen wrote WinCraps

[heavy]

Steen writes scripts for strategies he finds interesting. He's an International airline pilot who does a lot of his coding while on trans-oceanic flights. Plenty of dead-head time while the plane flies itself. That's why his software company is called Cloud City Software. Wonder if Covid 19 has him grounded or on limited flights these days or if he's still flying a regular schedule.

[Steen]

Steen writes scripts for strategies he finds interesting. He's an International airline pilot who does a lot of his coding while on trans-oceanic flights. Plenty of dead-head time while the plane flies itself. That's why his software company is called Cloud City Software. Wonder if Covid 19 has him grounded or on limited flights these days or if he's still flying a regular schedule.

Airlines are hit hard. I just flew yesterday for the first time in a month and won't fly gain for another two weeks! We'll see what happens once the economy is opened again, but it could take years for the airlines to recover.

Just to prevent any confusion, I do my programming on layovers or when riding as a passenger -- never when working in the cockpit.

I see there's some interest in a WinCraps script for Dave' System, so I took the liberty of writing and posting one. You can find it here:

https://cloudcitysoftware.com/Dave's System.bet

I based the script on the first posts from this thread. It's set to run 1000 sessions. Each session ends whenever $200 win goal is reached or bankroll drops below minimum bet. These values including bet amounts, bankroll, etc can easily be modified.

Of course you can run simulations with custom roll files or skewed probabilities. After running a simulation, results are best viewed on the Sessions Log but additional aggregated data can be found on various statistics screens by selecting "All sessions".

Enjoy!
Steen

[House of Orange]

Thanks Steen!