Basis for performance calculation

[botpro]

Hi,
in calculating the performance of a system, is it legitimate to use the average used capital instead of the initial capital?
Some background: my system works with "targets", but reaching the target varies by time (ie. sometime it takes 3 days, some other time 5 days etc., on average say it takes 4 days).
As part of the strategy it every time also uses a varying amount of capital from the avail cash in account, on average half of the initial capital.
Now, the problem is: which of the account values should be taken as the basis for the overall (ie. final) performance calculation:
1) the initial capital, or
2) the average used capital ?
The difference matters, because the avg used is only about half of the initial.

I'm making many runs (about 5 years), each single run (ie. each single period) starts afresh with the initial capital,
and then averaging the capital used, period duration, PL, etc. from all periods,
and then converting the averages to annual values (with and without compounding).
But the performance of course differs significantly between the to cases above.
Which one makes more sense? I tend to the 2nd case, but how to best argument/justify/convince?

 

[xandman]

Instead of analyzing the maths (ie time value calculations) and the various caveats, you may want to look up the standard performance measures used by the industry with regards to your specific type of strategy. Many people have spent quite a few brain cells identifying the best measures.

Who knows? You could be soliciting investors soon.

 

[botpro]

Instead of analyzing the maths (ie time value calculations) and the various caveats, you may want to look up the standard performance measures used by the industry with regards to your specific type of strategy. Many people have spent quite a few brain cells identifying the best measures.

Who knows? You could be soliciting investors soon.

Over the weekend I changed the code so that it now continously trades a full year (or longer).
This gives more realistic results by also using the profits in trading (ie. compounding), and saves me the previous hand calculations (actually did it in spreadsheet).
It now performs like in other backtesting programs, although with not that much bells and whistles.
The trading results look very good, I'll soon create a prospectus and invite possible investors.
(fyi: I've a good system but unfortunately no own funds to trade it myself, so I need to convince one or more investors of my system).

 

[newwurldmn]

Hi,
in calculating the performance of a system, is it legitimate to use the average used capital instead of the initial capital?
Some background: my system works with "targets", but reaching the target varies by time (ie. sometime it takes 3 days, some other time 5 days etc., on average say it takes 4 days).
As part of the strategy it every time also uses a varying amount of capital from the avail cash in account, on average half of the initial capital.
Now, the problem is: which of the account values should be taken as the basis for the overall (ie. final) performance calculation:
1) the initial capital, or
2) the average used capital ?
The difference matters, because the avg used is only about half of the initial.

I'm making many runs (about 5 years), each single run (ie. each single period) starts afresh with the initial capital,
and then averaging the capital used, period duration, PL, etc. from all periods,
and then converting the averages to annual values (with and without compounding).
But the performance of course differs significantly between the to cases above.
Which one makes more sense? I tend to the 2nd case, but how to best argument/justify/convince?

Typically you should use max capital (whatever the maximum required at any point is) otherwise the strategy is not employable in it's presented form. If you can't figure that out then often people will use a factor of VaR. Like 100x 99% daily var (meaning you can only lose >1% 1 out 100 days).

 

[botpro]

Typically you should use max capital (whatever the maximum required at any point is) otherwise the strategy is not employable in it's presented form. If you can't figure that out then often people will use a factor of VaR. Like 100x 99% daily var (meaning you can only lose >1% 1 out 100 days).

Thanks, I'll shortly post a sample output (just 3 months, ie. about 63 days = 63 lines), maybe then we could discuss which basis I better should take for the final PL calculation.

Ok here is the data:

Snapshot results at begin of day:

Day cOp OpVal OpPaid PL PLpct AV MinPosAmt nBots
1 0 0.00 0.00 0.00 0.00 600000.00 6000.00 25
2 21 126839.98 129622.94 7599.19 1.27 607599.19 6075.99 25
3 21 120724.54 126952.07 9462.13 1.58 609462.13 6094.62 25
4 22 123539.77 133798.38 12726.95 2.12 612726.95 6127.27 25
5 21 130490.53 139045.61 17737.60 2.96 617737.60 6177.38 25
6 20 142449.04 153207.00 21369.80 3.56 621369.80 6213.70 25
7 21 127745.20 130745.55 35780.81 5.96 635780.81 6357.81 25
8 21 139073.94 152429.00 28333.38 4.72 628333.38 6283.33 25
9 23 138106.44 144123.31 45787.22 7.63 645787.22 6457.87 25
10 23 167471.19 179950.05 44744.03 7.46 644744.03 6447.44 25
11 23 146465.43 149836.14 79682.52 13.28 679682.52 6796.83 25
12 23 146361.77 155144.83 82217.48 13.70 682217.48 6822.17 25
13 24 159529.89 163546.77 96317.10 16.05 696317.10 6963.17 25
14 23 158917.37 161864.21 107642.69 17.94 707642.69 7076.43 25
15 22 160748.68 165519.04 121376.56 20.23 721376.56 7213.77 25
16 22 157293.11 158966.92 143803.29 23.97 743803.29 7438.03 25
17 20 152692.98 162167.03 146755.14 24.46 746755.14 7467.55 25
18 21 170455.88 185463.43 150979.97 25.16 750979.97 7509.80 25
19 23 173734.39 182129.96 177505.67 29.58 777505.67 7775.06 25
20 20 177989.84 195350.10 180773.39 30.13 780773.39 7807.73 25
21 21 173821.97 175526.05 211600.46 35.27 811600.46 8116.00 25
22 23 183277.45 186629.28 227595.44 37.93 827595.44 8275.95 25
23 23 211250.00 243047.32 203987.99 34.00 803987.99 8039.88 25
24 18 150783.23 162558.52 233915.86 38.99 833915.86 8339.16 25
25 16 133052.51 133506.29 260915.34 43.49 860915.34 8609.15 25
26 20 177705.39 185589.54 265307.76 44.22 865307.76 8653.08 25
27 22 182552.23 190439.42 280876.06 46.81 880876.06 8808.76 25
28 20 167914.62 176883.71 295448.97 49.24 895448.97 8954.49 25
29 21 190949.12 207314.07 307159.04 51.19 907159.04 9071.59 25
30 21 224995.29 253928.10 302127.51 50.35 902127.51 9021.28 25
31 22 191177.96 191433.61 361255.34 60.21 961255.34 9612.55 25
32 22 198679.97 210181.78 361571.26 60.26 961571.26 9615.71 25
33 19 175109.99 179221.21 388493.44 64.75 988493.44 9884.93 25
34 21 192091.54 200853.89 397761.78 66.29 997761.78 9977.62 25
35 22 206395.93 212886.13 416533.60 69.42 1016533.60 10165.34 25
36 19 175348.20 193476.03 420909.72 70.15 1020909.72 10209.10 25
37 21 206217.82 212586.90 453249.39 75.54 1053249.39 10532.49 25
38 22 216434.77 234849.38 453700.91 75.62 1053700.91 10537.01 25
39 23 230775.76 247347.70 486666.95 81.11 1086666.95 10866.67 25
40 22 241562.64 259081.14 525860.95 87.64 1125860.95 11258.61 25
41 23 241998.10 249943.63 636310.44 106.05 1236310.44 12363.10 25
42 22 256800.33 265114.86 659849.28 109.97 1259849.28 12598.49 25
43 22 274446.98 290588.50 669391.09 111.57 1269391.09 12693.91 25
44 20 327610.93 362259.69 662173.99 110.36 1262173.99 12621.74 25
45 19 232979.91 256384.02 686886.51 114.48 1286886.51 12868.87 25
46 17 215427.87 222591.28 723460.45 120.58 1323460.45 13234.60 25
47 19 240157.94 247196.09 746083.16 124.35 1346083.16 13460.83 25
48 21 272923.80 277476.14 751920.69 125.32 1351920.69 13519.21 25
49 22 427072.15 464937.66 725624.31 120.94 1325624.31 13256.24 25
50 23 277661.11 300880.42 770287.36 128.38 1370287.36 13702.87 25
51 22 301376.55 318822.44 797271.41 132.88 1397271.41 13972.71 25
52 21 297245.87 306788.48 834631.72 139.11 1434631.72 14346.32 25
53 22 293623.18 308012.92 857330.29 142.89 1457330.29 14573.30 25
54 21 295629.24 313803.37 891663.24 148.61 1491663.24 14916.63 25
55 23 334102.12 344765.06 930437.84 155.07 1530437.84 15304.38 25
56 20 293264.24 297084.21 983074.52 163.85 1583074.52 15830.75 25
57 22 334249.50 353623.61 1012894.79 168.82 1612894.79 16128.95 25
58 21 329209.93 337281.56 1044465.44 174.08 1644465.44 16444.65 25
59 22 348246.97 370172.04 1060974.88 176.83 1660974.88 16609.75 25
60 21 335988.88 372295.65 1076734.27 179.46 1676734.27 16767.34 25
61 23 366292.46 389960.79 1127433.24 187.91 1727433.24 17274.33 25
62 22 360315.11 370897.44 1174783.07 195.80 1774783.07 17747.83 25
63 23 376223.42 400565.60 1172508.78 195.42 1772508.78 17725.09 25
64 19 330287.47 331647.81 1221909.81 203.65 1821909.81 18219.10 25

Initial account size is 600k, but as can be seen the system does not use all the avail funds, rather only a fraction of it.
So, which basis should be taken for the PL% calculation?
(above I used the initial 600k as basis, but this could be maybe an incorrect assumption)

PS: the above data is normally tabulated, but the posting system reformats it even though I enclosed it in a CODE block.