Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Hyperopt loss going wrong way ? #10192

Closed
taz4hvn opened this issue May 7, 2024 · 8 comments · Fixed by #10221
Closed

Hyperopt loss going wrong way ? #10192

taz4hvn opened this issue May 7, 2024 · 8 comments · Fixed by #10221
Labels
Hyperopt Hyperopt related issues and pull requests

Comments

@taz4hvn
Copy link

taz4hvn commented May 7, 2024

Describe your environment

  • Operating system: ____ Manjaro
  • Python Version: _____ Python 3.11.8
  • CCXT version: _____ ccxt==4.2.76
  • Freqtrade Version: ____ freqtrade 2024.4

Note: All issues other than enhancement requests will be closed without further comment if the above template is deleted or not filled out.

Describe the problem:

Hi. It seems loss function score is broken somehow (maybe smthg to do with its sign).
Example below (as on 2024-05-07) says #5 is best while obviously it should be #10.
Regards.

Steps to reproduce:

  1. command used:
    docker-compose run --rm freqtrade hyperopt --disable-param-export --config ./user_data/config_binanceWEB.json --strategy TAZ_15m_binance000 --timerange=20230901-20240501 --spaces buy --hyperopt-loss ProfitDrawDownHyperOptLoss --epoch 1000 --timeframe 15m --timeframe-detail 1m --j 1

Observed Results:

+--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------+
|   Best |   Epoch |   Trades |    Win  Draw  Loss  Win% |   Avg profit |                    Profit |   Avg duration |    Objective |       Max Drawdown (Acct) |
|--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------|
| * Best |   1/135 |        4 |      3     0     1  75.0 |        1.59% |        30.785     (3.08%) |        1:35:00 |    -29.70234 |         3.559     (0.35%) |
| *      |   2/135 |       11 |      8     0     3  72.7 |       -0.94% |      -221.333   (-22.13%) |        0:24:00 |    247.98383 |       145.339    (15.98%) |
| *      |   3/135 |       51 |     42     0     9  82.4 |        0.43% |      -315.876   (-31.59%) |        0:54:00 |     87.71371 |       417.773    (40.04%) |
| *      |   4/135 |        1 |      1     0     0   100 |        2.29% |         3.795     (0.38%) |        0:00:00 |     -3.79499 |                        -- |
| * Best |   5/135 |       51 |     42     0     9  82.4 |        0.11% |      -459.923   (-45.99%) |        0:28:00 | -4,180.04668 |       510.315    (50.84%) |
| *      |   6/135 |      170 |    145     0    25  85.3 |        0.69% |      -581.544   (-58.15%) |        0:52:00 |   -100.19843 |       721.638    (68.98%) |
| *      |   7/135 |        2 |      2     0     0   100 |        2.29% |         7.604     (0.76%) |        0:00:00 |     -7.60435 |                        -- |
| *      |   8/135 |       15 |     14     0     1  93.3 |        1.78% |        64.167     (6.42%) |        1:46:00 |    -59.80514 |        56.433     (5.31%) |
| *      |   9/135 |      184 |    158     0    26  85.9 |        0.74% |      -628.566   (-62.86%) |        0:46:00 |     11.19232 |       790.266    (74.53%) |
| *      |  10/135 |      190 |    174     0    16  91.6 |        1.48% |       602.105    (60.21%) |        1:01:00 |   -586.82945 |       304.910    (16.04%) |
| *      |  11/135 |        5 |      5     0     0   100 |        2.29% |        22.971     (2.30%) |        0:03:00 |    -22.97094 |                        -- |
| *      |  12/135 |      345 |    298     0    47  86.4 |        0.94% |      -494.568   (-49.46%) |        1:21:00 |     17.80432 |       746.868    (70.59%) |
| *      |  13/135 |       44 |     37     0     7  84.1 |        0.64% |      -251.262   (-25.13%) |        0:55:00 |   -624.95887 |       352.895    (35.02%) |
| *      |  14/135 |        2 |      2     0     0   100 |        2.29% |         7.604     (0.76%) |        0:00:00 |     -7.60435 |                        -- |
| *      |  15/135 |        4 |      4     0     0   100 |        2.29% |        15.263     (1.53%) |        0:01:00 |    -15.26296 |                        -- |
| *      |  16/135 |       49 |     40     0     9  81.6 |        0.20% |      -429.174   (-42.92%) |        0:35:00 |   -901.90677 |       471.865    (46.65%) |
| *      |  17/135 |      362 |    310     0    52  85.6 |        0.79% |      -759.062   (-75.91%) |        1:10:00 | -1,031.79724 |       842.359    (82.04%) |
@taz4hvn taz4hvn added the Triage Needed Issues yet to verify label May 7, 2024
@xmatthias xmatthias added Hyperopt Hyperopt related issues and pull requests and removed Triage Needed Issues yet to verify labels May 7, 2024
@xmatthias
Copy link
Member

Well the calculation is not really ideal - that's also been reported in #9563 and #9819.

The sign is however not "really" the problem (at least not the only one) - it's the overall combination in the formula - as well as the (at least in part, ignorance of the range of values a value can reach.

While the variable says "max_drawdown_abs" - it's really a relative value that's retrieved.
it's however NOT bound to be between 0 and one - but rather, it's unbound in it's upper bound, as the account level drawdown can be > 100% (it's drawdown from the high point, not from starting balance).

So if your strategy does great, your "high" value may be 100% profit (from 1000$ to 2000$) - but then it's in a bad run and loses 1200$ (your net-loss is 200$).
What you'd get then is ~1.2 (120%).


for your particular result - if you simply invert the sign - you'd get #5 as "worst" result ... but #2 as best ... which is still a rather losing strategy.
#12 would be pretty high up, too - which is losing AND has a pretty high drawdown - both not very ideal situations.


In reality / summary - i think the way this loss function combines the values is not ideal - and I'm more than happy to get a PR that sees it improved.

The (simple, from a user perspective) alternative is to pick a different loss function - which there's plenty of - and all try to search for different "improvements".

@taz4hvn
Copy link
Author

taz4hvn commented May 12, 2024

Well, thanks for the comprehensive answer. I guess I'll stick with OnlyProfitHyperOptLoss and manually discard frightening drawdowns for now.

@taz4hvn
Copy link
Author

taz4hvn commented May 23, 2024

Hi. I tested 2024.5-dev-ea27a1ec implementing new formula proposed by #10221. Now objective is positive so freqtrade actually tries to improve the worst winning strategy. Appears to me a simple minus sign or min/max swap is needed.

Example:

+--------+---------+----------+--------------------------+--------------+---------------------------+----------------+---------------+---------------------------+
| Best | Epoch | Trades | Win Draw Loss Win% | Avg profit | Profit | Avg duration | Objective | Max Drawdown (Acct) |
|--------+---------+----------+--------------------------+--------------+---------------------------+----------------+---------------+---------------------------|
| * | 1/76 | 734 | 602 0 132 82.0 | 3.95% | 19556.873 (1,955.69%) | 3:55:00 | 38,257.15248 | 4139.799 (22.17%) |
| * | 2/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 3/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 4/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 5/76 | 1012 | 791 0 221 78.2 | 3.21% | 20645.571 (2,064.56%) | 4:19:00 | 66,862.57108 | 9622.858 (31.79%) |
| * | 6/76 | 601 | 498 0 103 82.9 | 4.39% | 14163.363 (1,416.34%) | 4:01:00 | 28,566.10630 | 3043.671 (22.63%) |
| * | 7/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 8/76 | 599 | 495 0 104 82.6 | 4.31% | 12034.538 (1,203.45%) | 4:02:00 | 24,272.50466 | 2627.944 (22.63%) |
| * | 9/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 10/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 11/76 | 629 | 517 0 112 82.2 | 4.30% | 17080.542 (1,708.05%) | 4:07:00 | 34,070.72703 | 3684.872 (22.46%) |
| * | 12/76 | 744 | 609 0 135 81.9 | 3.87% | 19631.124 (1,963.11%) | 3:52:00 | 38,402.39922 | 4246.497 (22.17%) |
| * | 13/76 | 684 | 558 0 126 81.6 | 4.11% | 16423.782 (1,642.38%) | 3:59:00 | 33,935.62496 | 3942.400 (23.00%) |
| * | 14/76 | 629 | 522 0 107 83.0 | 4.34% | 15421.954 (1,542.20%) | 3:59:00 | 32,985.99514 | 3466.805 (23.54%) |
| * | 15/76 | 602 | 498 0 104 82.7 | 4.35% | 13823.615 (1,382.36%) | 4:01:00 | 28,923.84756 | 3085.955 (23.19%) |
| * | 16/76 | 726 | 594 0 132 81.8 | 3.91% | 19059.908 (1,905.99%) | 3:45:00 | 41,485.29107 | 4725.894 (23.82%) |
| * | 17/76 | 598 | 495 0 103 82.8 | 4.35% | 13328.017 (1,332.80%) | 4:02:00 | 27,886.88265 | 2982.782 (23.19%) |
| * | 18/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 19/76 | 631 | 523 0 108 82.9 | 4.24% | 14924.618 (1,492.46%) | 3:55:00 | 31,303.92590 | 3679.231 (23.23%) |
| * | 20/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 21/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 22/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 23/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 24/76 | 599 | 496 0 103 82.8 | 4.37% | 13702.669 (1,370.27%) | 4:02:00 | 28,670.80167 | 3025.407 (23.19%) |
| * | 25/76 | 606 | 500 0 106 82.5 | 4.32% | 12287.715 (1,228.77%) | 4:07:00 | 24,783.12308 | 2652.318 (22.63%) |
| * | 26/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 27/76 | 609 | 502 0 107 82.4 | 4.34% | 14710.133 (1,471.01%) | 4:01:00 | 30,778.73363 | 3239.703 (23.19%) |
| * | 28/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 29/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| * | 30/76 | 1129 | 896 0 233 79.4 | 3.08% | 36618.966 (3,661.90%) | 3:50:00 | 130,356.35571 | 18148.476 (34.20%) |
| | 31/76 | 795 | 651 0 144 81.9 | 3.55% | 15774.820 (1,577.48%) | 3:33:00 | 33,186.70003 | 4172.133 (23.28%) |
| | 32/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| | 33/76 | 599 | 496 0 103 82.8 | 4.35% | 13413.411 (1,341.34%) | 4:01:00 | 27,053.54215 | 2905.942 (22.63%) |
| | 34/76 | 598 | 494 0 104 82.6 | 4.31% | 11937.076 (1,193.71%) | 4:02:00 | 24,976.56314 | 2693.220 (23.19%) |
| | 35/76 | 601 | 497 0 104 82.7 | 4.33% | 12515.382 (1,251.54%) | 4:01:00 | 26,186.56351 | 2801.211 (23.19%) |
| | 36/76 | 601 | 497 0 104 82.7 | 4.34% | 12499.879 (1,249.99%) | 4:01:00 | 26,154.14337 | 2810.383 (23.19%) |
| | 37/76 | 597 | 493 0 104 82.6 | 4.30% | 11652.598 (1,165.26%) | 4:02:00 | 24,381.33363 | 2633.998 (23.19%) |
| | 38/76 | 600 | 496 0 104 82.7 | 4.32% | 12306.063 (1,230.61%) | 4:01:00 | 24,820.12574 | 2667.735 (22.63%) |
| | 39/76 | 600 | 496 0 104 82.7 | 4.33% | 12377.952 (1,237.80%) | 4:01:00 | 24,965.13560 | 2682.150 (22.63%) |
| | 40/76 | 607 | 502 0 105 82.7 | 4.36% | 13275.814 (1,327.58%) | 3:59:00 | 26,928.87442 | 2863.620 (22.71%) |
| | 41/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| | 42/76 | 599 | 495 0 104 82.6 | 4.30% | 12156.656 (1,215.67%) | 4:02:00 | 25,435.98772 | 2738.930 (23.19%) |
| | 43/76 | 601 | 497 0 104 82.7 | 4.31% | 12291.424 (1,229.14%) | 4:01:00 | 25,858.44901 | 2770.979 (23.28%) |
| | 44/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| | 45/76 | 603 | 499 0 104 82.8 | 4.34% | 12759.322 (1,275.93%) | 4:00:00 | 25,881.21752 | 2772.232 (22.71%) |
| | 46/76 | 598 | 494 0 104 82.6 | 4.29% | 11750.330 (1,175.03%) | 4:02:00 | 24,585.83273 | 2654.345 (23.19%) |
| | 47/76 | 599 | 495 0 104 82.6 | 4.30% | 12039.121 (1,203.91%) | 4:02:00 | 25,190.06864 | 2714.462 (23.19%) |
| | 48/76 | 600 | 496 0 104 82.7 | 4.31% | 12795.640 (1,279.56%) | 4:01:00 | 25,807.56056 | 2781.391 (22.63%) |
| | 49/76 | 599 | 495 0 104 82.6 | 4.30% | 12098.794 (1,209.88%) | 4:02:00 | 25,314.91856 | 2726.884 (23.19%) |
| | 50/76 | 605 | 497 0 108 82.1 | 4.28% | 12721.381 (1,272.14%) | 4:00:00 | 25,804.24734 | 2855.773 (22.71%) |
| | 51/76 | 602 | 498 0 104 82.7 | 4.35% | 12664.180 (1,266.42%) | 4:00:00 | 25,542.43362 | 2739.536 (22.63%) |
| | 52/76 | 604 | 500 0 104 82.8 | 4.32% | 13275.071 (1,327.51%) | 4:00:00 | 27,927.83235 | 2986.077 (23.28%) |
| | 53/76 | 604 | 499 0 105 82.6 | 4.34% | 12950.335 (1,295.03%) | 3:58:00 | 26,268.66078 | 2801.277 (22.71%) |
| | 54/76 | 604 | 499 0 105 82.6 | 4.32% | 12934.186 (1,293.42%) | 4:00:00 | 26,235.90012 | 2923.139 (22.71%) |
| | 55/76 | 602 | 498 0 104 82.7 | 4.33% | 12849.671 (1,284.97%) | 4:00:00 | 25,916.52942 | 2776.723 (22.63%) |
| | 56/76 | 603 | 499 0 104 82.8 | 4.31% | 13075.987 (1,307.60%) | 4:01:00 | 26,372.97766 | 2850.563 (22.63%) |
| | 57/76 | 599 | 495 0 104 82.6 | 4.30% | 12823.897 (1,282.39%) | 4:02:00 | 26,832.09730 | 2877.837 (23.19%) |
| | 58/76 | 602 | 498 0 104 82.7 | 4.31% | 12975.882 (1,297.59%) | 4:01:00 | 26,171.09430 | 2802.028 (22.63%) |
| | 59/76 | 604 | 500 0 104 82.8 | 4.36% | 13023.474 (1,302.35%) | 4:00:00 | 26,267.07810 | 2811.570 (22.63%) |
| | 60/76 | 605 | 500 0 105 82.6 | 4.35% | 13089.809 (1,308.98%) | 4:00:00 | 26,400.87714 | 2812.422 (22.63%) |
| | 61/76 | 602 | 498 0 104 82.7 | 4.31% | 12975.882 (1,297.59%) | 4:01:00 | 26,171.09430 | 2802.028 (22.63%) |
| | 62/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| | 63/76 | 607 | 502 0 105 82.7 | 4.30% | 12297.110 (1,229.71%) | 4:00:00 | 24,943.64966 | 2781.250 (22.71%) |
| | 64/76 | 613 | 509 0 104 83.0 | 4.34% | 14645.958 (1,464.60%) | 3:57:00 | 30,811.86288 | 3223.874 (23.28%) |
| | 65/76 | 602 | 498 0 104 82.7 | 4.32% | 13356.179 (1,335.62%) | 4:01:00 | 26,938.09286 | 2878.272 (22.63%) |
| | 66/76 | 604 | 500 0 104 82.8 | 4.36% | 13023.474 (1,302.35%) | 4:00:00 | 26,267.07810 | 2811.570 (22.63%) |
| | 67/76 | 616 | 511 0 105 83.0 | 4.30% | 14003.897 (1,400.39%) | 3:57:00 | 28,405.72266 | 3090.325 (22.71%) |
| | 68/76 | 598 | 495 0 103 82.8 | 4.35% | 13307.999 (1,330.80%) | 4:02:00 | 27,845.00588 | 2978.616 (23.19%) |
| | 69/76 | 599 | 495 0 104 82.6 | 4.31% | 12456.486 (1,245.65%) | 4:02:00 | 24,847.12094 | 2679.339 (22.46%) |
| | 70/76 | 605 | 499 0 106 82.5 | 4.29% | 12492.518 (1,249.25%) | 4:06:00 | 25,349.64940 | 2726.529 (22.72%) |
| | 71/76 | 600 | 496 0 104 82.7 | 4.31% | 12591.008 (1,259.10%) | 4:02:00 | 25,115.43947 | 2706.122 (22.46%) |
| | 72/76 | 611 | 504 0 107 82.5 | 4.25% | 13014.801 (1,301.48%) | 4:05:00 | 26,409.46781 | 2832.072 (22.72%) |
| | 73/76 | 603 | 499 0 104 82.8 | 4.33% | 13473.301 (1,347.33%) | 4:01:00 | 28,190.86992 | 3013.028 (23.19%) |
| | 74/76 | 613 | 508 0 105 82.9 | 4.29% | 13777.477 (1,377.75%) | 4:01:00 | 27,787.81039 | 2961.171 (22.63%) |
| | 75/76 | 598 | 494 0 104 82.6 | 4.31% | 11937.076 (1,193.71%) | 4:02:00 | 24,976.56314 | 2693.220 (23.19%) |
| | 76/76 | 610 | 503 0 107 82.5 | 4.26% | 12962.665 (1,296.27%) | 4:01:00 | 26,144.41571 | 2903.681 (22.63%) |
+--------+---------+----------+--------------------------+--------------+---------------------------+----------------+---------------+---------------------------+

PS: I dunno if commenting on this post was the correct procedure ?

@xmatthias
Copy link
Member

It's interesting insofar as my experience was different - and results DID improve quite significantly, preferring better results.

@xmatthias
Copy link
Member

Are you sure that you're using the correct loss function (ProfitDrawDownHyperOptLoss)

I'm asking this because your objective's do not align with that calculation.
Using your very first result - the calculation happens with the following parameters

DRAWDOWN_MULT = 0.075
total_profit  = 19556
rel_drawdown = 22.17 / 100

-1 * (total_profit - (rel_drawdown * DRAWDOWN_MULT * total_profit))

and results in an objective of -19230.83261.

Your objective on the other hand is 38,257.15248 - which is something completely different - and i don't really see how this can happen with the given values.

@taz4hvn
Copy link
Author

taz4hvn commented May 24, 2024

hi,

FIRST RUN

I did a new test, since previous is lost.

COMMAND:
docker-compose run --rm freqtrade hyperopt --disable-param-export --config ./user_data/config_kucoinWEB.json --strategy TAZ_kucoin320b --timerange=20231201-20240522 --spaces buy --hyperopt-loss ProfitDrawDownHyperOptLoss --epoch 1000 --timeframe 5m --timeframe-detail 1m --j 1

FREQTRADE:
freqtrade docker-2024.5-dev-ea27a1ec

OUTPUT:

+--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------+
|   Best |   Epoch |   Trades |    Win  Draw  Loss  Win% |   Avg profit |                    Profit |   Avg duration |    Objective |       Max Drawdown (Acct) |
|--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------|
| *      |     1/2 |      985 |    806     0   179  81.8 |        3.57% |    39529.212  (3,952.92%) |        3:39:00 | 76,461.97295 |      7606.080    (22.01%) |
| *      |     2/2 |     1095 |    878     0   217  80.2 |        3.40% |    49276.039  (4,927.60%) |        4:05:00 | 99,701.15153 |     10293.676    (22.67%) |
+--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------+

CALCULATION:
for epoch 1: -1x(39529.212-(0.2201x0.075x39529.212)=-38876 same weird behavior you pointed.

SECOND RUN

after a pull

FREQTRADE:
freqtrade docker-2024.5-dev-2c740059

OUTPUT:

+--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------+
|   Best |   Epoch |   Trades |    Win  Draw  Loss  Win% |   Avg profit |                    Profit |   Avg duration |    Objective |       Max Drawdown (Acct) |
|--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------|
| *      |     1/2 |      986 |    807     0   179  81.8 |        3.58% |    39887.289  (3,988.73%) |        3:39:00 | 77,154.60589 |      7673.281    (22.01%) |
| *      |     2/2 |     1020 |    836     0   184  82.0 |        3.59% |    48260.136  (4,826.01%) |        3:38:00 | 93,350.30176 |      9424.433    (22.01%) |
+--------+---------+----------+--------------------------+--------------+---------------------------+----------------+--------------+---------------------------+

same positive Objective values with this build.

@taz4hvn
Copy link
Author

taz4hvn commented May 24, 2024

in:
https://github.com/freqtrade/freqtrade/blob/develop/freqtrade/optimize/hyperopt_loss/hyperopt_loss_profit_drawdown.py
i can read:
return -1 * (total_profit - (relative_account_drawdown * total_profit) / DRAWDOWN_MULT)
when applied to my last first run gives 76475 for epoch1 and 99669 for epoch2, pretty close (but not equal ?) to what I got.
Seems formula is not what you think it is ...

PS: coherent with 480477d

@xmatthias
Copy link
Member

You're right, the one i used in my calculations was not the very last try.
i've updated this now to how i ACTUALLY wanted to commit it - so it should work better now.

xmatthias added a commit that referenced this issue May 25, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Hyperopt Hyperopt related issues and pull requests
Projects
None yet
2 participants