Contact

Precision Stop (6-12) Multiple test on huge data set
Testing of Precision stop (6-12) Para multiple on 123 various stock, I have chosen only a few good system stocks, the rest are chosen because they were known to me. Ten years data tested. Test run from Jan 3rd 2000 to August 12th 2010 These tests used the same raw database and copied and pasted it and ran indicator builder over it each time. Candover Investments, JKX oil and gas, Quintain estates, Pendragon, JJB sports, Johnson press, Royal bank of Scotland, Lloyds TSB, Unite group, Polar capital tech trust. and various others...see excel list of stocks used Before the test, I predict 0.83 multiple will work best....lets see. Future upgrades Remove opacity feature Luxury future items. Make the 6-12 averages changeable Make the noise look back ( currently 14 days) up to a higher number to add flexibility to the stop. Test volatility break out system Test Mach Trend system Purpose of this test is to ascertain the best GLOBAL multiple settings in a simulation run over realistic portfolio over a large amount of sample. Settings used are below as there is a good mixture of small and large stocks, the 1.5 per side spread setting allows a harsh setting which errs on the "over strict" the slippage amount of 50% also errs on the over strict side. So it might be noted that the optimal stop setting found might end up being slightly faster than those shown in the simulation. But it will be reasonably close. This test used the new version of the simulation unit which has the high and lows used to trigger the stops, and hence has higher accuracy. The risk profile is set to a slightly conservative 0.25 of capital risked to stops The trade divisor which divides up the risk ( £10,000 x 0.25 ) = £2,500 then divided by 80 = £31.25 risk per trade x 123 = as many symbols have no data in the initial part of the test, a slightly smaller divisor than 123 will be used. I have chosen 80 The max cost of one trade will be set to 4.0% of equity so initially we have £400 deal sizes. This maxes out at 123 x 400 = £49,299 = 4.9x gearing The gearing will have a restriction placed on it at 7 times in case a large draw down raises the gearing over 4.5 which is the intended max.
Test no	Para multiple	Risk %	Min per	Max per	Max cost %	Trade divisor	Max trades day	Spread per side	Max gear (Limited)	Slippage	Starting equity
1	1	0.25	0.01	0.20	4.00%	80	123	1.5 (3% total)	7	50%	£10,000
2	1	0.25	0.01	0.20	4.25%	75	123	1.5	7	50%	£10,000
3	1	0.25	0.01	0.20	4.50%	70	123	1.5	7	50%	£10,000
4	1	0.25	0.01	0.20	4.70%	65	123	1.5	7	50%	£10,000
5	0.9	0.25	0.01	0.20	4.70%	65	123	1.5	7	50%	£10,000
6	0.8	0.25	0.01	0.20	4.70%	65	123	1.5	7	50%	£10,000
7	0.7	0.25	0.01	0.20	4.70%	65	123	1.5	7	50%	£10,000
7.1	0.7	0.25	0.01	0.20	4.50%	68	123	1.5	7	50%	£10,000
8	0.6	0.25	0.01	0.20	4.50%	68	123	1.5	7	50%	£10,000
8.1	0.6	0.25	0.01	0.20	4.40%	70	123	1.5	7	50%	£10,000
9	0.5	0.25	0.01	0.20	4.40%	70	123	1.5	7	50%	£10,000
9.1	0.5	0.25	0.01	0.20	4.25%	75	123	1.5	7	50%	£10,000
9.2	0.5	0.25	0.01	0.20	4.10%	80	123	1.5	7	50%	£10,000
10	0.4	0.25	0.01	0.20	4.10%	80	123	1.5	7	50%	£10,000
10.1	0.4	0.25	0.01	0.20	4.00%	85	123	1.5	7	50%	£10,000
10.2	0.4	0.25	0.01	0.20	3.90%	90	123	1.5	7	50%	£10,000
11	0.3	0.25	0.01	0.20	3.80%	95	123	1.5	7	50%	£10,000
				Testing longer term setting below 1 to 2 multiple
12	1.2	0.25	0.01	0.20	4.80%	65	123	1.5	7	50%	£10,000
12.1	1.2	0.25	0.01	0.20	4.90%	60	123	1.5	7	50%	£10,000
13	1.4	0.25	0.01	0.20	4.90%	60	123	1.5	7	50%	£10,000
14	1.6	0.25	0.01	0.20	5%	55	123	1.5	7	50%	£10,000
14.1	1.6	0.25	0.01	0.20	5.25%	50	123	1.5	7	50%	£10,000
15	1.8	0.25	0.01	0.50	5.25%	50	123	1.5	7	50%	£10,000
					Below is fine tuning of Multiple after a brief reflection
	.58
23	.56	0.25	.01	.2	4.2%	74	123	1.5	7	50%	10,000
16	.55	0.25	.01	.2	4.4%	72	123	1.5	7	50%	10,000
16.1	.55	0.25	.01	.2	4.2%	74	123	1.5	7	50%	10,000
22	.54	0.25	.01	.2	4.1	75	123	1.5	7	50%	10,000
17	.52	0.25	.01	.2	4.1%	80	123	1.5	7	50%	10,000
18	.48	0.25	.01	.2	4%	82	123	1.5	5	50%	10,000
19	.44	0.25	.01	.2	3.9%	83	123	1.5	5	50%	10,000
20	.46	0.25	.01	.2	3.97%	83	123	1.5	5	50%	10,000
21	.42	0.25	.01	.2	3.95%	84	123	1.5	5	50%	10,000

Test No.	Para Multiple	Percent profit	Final Equity	Max DD%	U-W days	K	Win Lose Ratio	R/R Ratio	Total trades	Highest Gearing	Comments

1	1.0	67,296%	£6,739,607	-14.48	187	0.0524	.4099	1.661	9871	3.779	Serious test, showing good results, underwater is bad. Gearing never got above 4 so trade divisor can be reduced and max cost can be increased in test 2
2	1.0	96,121%	£9,622,156	-15.33	187	.0478	.4099	1.640	9871	4.032	Again gearing is under the 4.5 target, so shall increase the max cost% and trade divisor again for test 3
3	1.0	140,205%	£14,030,508	-16.30	187	.0426	.4099	1.617	9871	4.284	Again gearing is under the 4.5 target, so shall increase the max cost% and trade divisor again for test 4
4	1.0	201,402%	£20,150,267	-17.43	187	.0368	.4099	1.592	9871	4.52	Gearing is now at the required level. U/w and DD are a little high, so will try to reduce this. How?

Now the correct level of gearing is found, testing of different Multiple settings can begin. Have chosen 0.9 multiple as this may reduce draw downs
5	0.9	244,484%	£24,458,453	-17.68	148	.0291	.4062	1.585	11301	4.554	Huge improvement in profit. Slight improvement in UW
6	0.8	303,389%	£30,348,966	-19.67	150	.0291	.4062	1.585	13,301	4.649	Slight improvement in profit. Max dd + UW worse
7	0.7	436,883%	£43,698,346	-16.95	124	.042	.3963	1.708	16,076	4.858	Gearing over high, will retest with different divisor and max percent trade, Increased profit, good K, reduced UW and DD, good result
7.1	0.7	322,410%	£32,251,019	-16.31	124	.0441	.3963	1.718	16,076	4.644	Gearing was too high in last test, so have reduced max percent trade to 4.50% and divisor up to 68 from 65. Notice profit has dropped 11million from a 0.2 reduction in gearing
8	0.6	1,032,935%	£103,303,542	-21.36	183	.0404	.394	1.716	20381	4.73	Gearing again rising with faster system? Testing again with smaller bets...huge profits
8.1	0.6	857,814,%	£85,791,411	-20.93	183	.0417	.394	1.7224	20381	4.62	Gearing ok? Suprised to see faster system doing so well?
9	0.5	2,704,541%	£270,464,174	-19.76	298	.041	.3843	1.794	26985	4.941	Again gearing is rising with speed. Need to cut this to get fair comparison (see test 8.1)
9.1	0.5	1,852,190%	£185,229,060	-19.18	298	.0435	.3843	1.807	26985	4.761	Again gearing is rising with speed. Need to cut this to get fair comparison (see test 8.1)
9.2	0.5	1,280,775%	£128,087,542	-18.60	298	.0459	.3843	1.820	26985	4.584	Gearing is now ok but differences in profit are phenomical with small gearing changes (see test 9 to 9.2)
10	0.4	1,733,123%	£173,322,354	-38.39	981	.0823	.3377	2.118	35910	4.779	Gearing is rising as frequency is reduced. Incredibly this busy model is still making huge gains (see test 10.1)
10.1	0.4	1,320,832%	£132,093,207	-37.89	981	.0823	.3377	2.112	35910	4.653	Gearing is rising as frequency is reduced. Incredibly this busy model is still making huge gains (see test 10.1)
10.2	0.4	1,000,921%	£100,902,184	-37.28	981	.0851	.3374	2.132	35910	4.53	Gearing ok now DD and underwater are sucking
11	0.3	309,094%	£30,919,406	-66.09	1186	.0118	.3714	2.491	44086	4.54	Finally the expected decline in performance from over busy system is seen. DD and UW makes this setting not useable

Initial conclusions.

Target gearing has been around 4.5 times equity, but the above tests are not exact. During these tests the highest gearing only shows during the last few years of the test when the full data set is active.

Settings used are revealing a surprisingly high frequency system to be working best. The overall impression shows best values lie between 0.7 and 0. 4 with the likely best being between 0.45 and 0.55.

Trading with this set up the system generates 20,000 to 30,000 trades over a 10 year period which on 123 stocks equates to around 2,500 trades per years / 250 = 10 trades per day.

This seems a higher frequency than I originally expected.

Paradoxically the best profit settings between 0.4 and 0.6 gave very poor underwater readings.

What to do next? Mull over the results and go in the direction which can produce a model that I can stick to. Enduring more than 200 days underwater is not something I wish to do and neither do I want to have more than 20% drawdowns.

Basis of this 0.6 to 0.7 seem to look favourable. Next step is to test in 0.025 increments over the best area and see what comes up.

It would also be wise to test 1.25, 1.5, 2.0 and 2.5 just to be sure there isn't a double top shaped distribution curve hiding on the chart below.

Hence, now testing 1.2, 1.4, 1.6 etc if no improvements after that will abort long term testing. I am not expecting great results from the slow model.

last test was 81 divisor and 4.05% max trade, and had 4.53 gearing, so its still too high

	Mult	percent	profit	dd	uw	k	Winlose	R-R	trade	gear
12	1.2	111,466%	£11,156,617	-19.14	194	.0613	.4092	1.712	8095	4.389	Gearing too low, increase risk...
12.1	1.2	154,981%	£15,508,166	-20.45	195	.0551	.4092	1.682	8095	4.631	Gearing close enough
13	1.4	67,684%	£6,778,406	-21.07	195	.0501	.4030	1.704	7109	4.402	Gearing close enough
14	1.6	68,080%	£6,897,997	-26.39	265	.0442	.4026	1.688	6456	4.32	Gearing low so test again with more risk...
14.1	1.6	106,718%	£10,681,865	-28.43	265	.0442	.4026	1.649	6456	4.69	Gearing slightly over. but nothing special in this test.
14.1	1.6	106,718%	£10,681,865	-28.43	265	.0442	.4026	1.649	6456	4.69	At this point it seems wise to increase stop dist max per as stops would likely be getting limited by the 20% rule. lets see in next test.
15	1.8	3101%	£320,117	-18.63	290	.0125	.4104	2.12	2945	3.811	Have increased stop dist max per to 50% as stops would likely be getting limited by the 20% rule. lets see in next test.
				Fine tuning tests below, more accuracy on the sweet spot



16	0.55	1,368,464%	£136,856.404	-23.39	297	.0399	0.388	1.758	11,722	4.7	Gearing over, but profit more than with 0.6 multi, retest with less gear
16.1	0.55	991,128%	£99,122,827	-22.54	297	.0415	0.388	1.766	11,722	4.521	Gearing ok, but 0.5 is still top score, will test 0.52 next
17	0.52	948,238%	£94,883,830	-19.03	199	.0506	0.385	1.839	12,897	4.511	Noticeable drop in underwater and drawdown which is nice
18	0.48	1,003,831%	£100,393,141+G	-21.27	298	.0469	0.383	1.831	28,594	4.491	Gearing slightly under, but not worth retesting as its close
20	0.46	1,042,607%	£104,270,696	-23.75	314	.0554	0.380	1.903	30,400	4.497	Still not beating 0.5 but remarkable clear distribution of results
19	0.44	975,228%	£97,532,799	-22.37	313	.0627	0.378	1.968	32,144	4.506	Was expecting this one to be better, now 0.5 seems to stand tall
21	0.42	1,084,757%	£108,485,785-G	-25.97	947	.0708	0.377	2.030	34,034	4.597	UW goes through the roof here.
22	0.54	930,688%	£93,078,886	-21.77	298	.0448	0.387	1.785	24,096	4.508	Same same, 0.5 still is tops
23	0.56	823,807%	£82,390,712	-21.09	189	.091	0.388	1.764	22,786	4.513	Improved underwater, but less profit.

Testing longer term model shows performance to be severely impaired. Next testing will be on the minor multiple settings between 0.4 and 0.8 which is the sweetspot. Below chart shows the latest tests in green and the previous coarser increments in blue, I am expecting the best setting to be between 0.44 and 0.56

Vertical axis = Millions x 10 Horizontal axis = Multiple setting
13
12
11
10
9
8
7
6
5
4
3
2
1
1
	1.8	1.6	1.4	1.2	1.0	0.9	0.8	0.7	0.6	0.5	0.4	0.3

Vertical axis = Millions x 10 FINE TUNING TEST (New tests in green) Horizontal axis = Multiple setting
18
17
16
15
14
13
12
11
11
10
10
9
8
7
6
5
4
3
2
1
1
	.34	.36	.38	.40	.42	.44	.46	.48	.50	.52	.54	.55	.56	.58	.60	.62	.64