Skip to content

Add debug to pt_traininer and a markdown file to demonstrate the training processing findings #29

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.

Sign up for GitHub

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

Open
crispyduck91952 wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Add debug to pt_traininer and a markdown file to demonstrate the training processing findings #29

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
49322b6
Added debug to pt_traininer and a markdown file with a description of…
Jan 20, 2026
6942eda
Added detailed notes on the issue
Jan 20, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,9 @@
# This fork is only to explore if the training process is working as expected

Please read the debug notes here [debug_original_pt_trainer_notes.md](./debug_original_pt_trainer_notes.md).

The original README.md is below:

# PowerTrader_AI
Fully automated crypto trading powered by a custom price prediction AI and a structured/tiered DCA system.

Expand Down
81 changes: 81 additions & 0 deletions debug_original_pt_trainer_notes.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,81 @@
# Purpose

This relates to a clarification question related to the training process documented in Issue #27 here: https://github.com/garagesteve1155/PowerTrader_AI/issues/27

The following is an extract of the console output from pt_trainer training filtered by lines containing "[DEBUG]" whilst training the BTC coin.

# Observations

```
1. The 1hour timeframe trains on the same 25% of prices two times (restart count <= 2), then on restart count 3, it processes 100% of the 1hour prices.

2. The trainer processes 7 timeframes (see seven groups of output below), but they are not the timeframes you expect.

For example:
- group 1 process 25% of 1hour, then 25% of 1hour then all of 1hour
- group 2 process 25% of 1hour, then 25% of 2hour then all of 2hour
- group 3 process 25% of 1hour, then 25% of 4hour then all of 4hour
- group 4 process 25% of 1hour, then 25% of 8hour then all of 8hour
- group 5 process 25% of 1hour, then 25% of 12hour then all of 12hour
- group 6 process 25% of 1hour, then 25% of 1day then all of 1day
- group 7 process 25% of 1hour, then 25% of 1week then all of 1week
```

### The clarification is to check the below is what the author expected as the design doesn't seem correct.

```
[DEBUG] Training timeframe 1hour, restart count:0
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 1hour, restart count:1
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 1hour, restart count:2
[DEBUG] Coin BTC-GBP: 100% of 13121 prices for timeframe 1hour have been processed. Saving training status then exiting.


[DEBUG] Training timeframe 1hour, restart count:0
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 2hour, restart count:1
[DEBUG] Coin BTC-GBP: 25.0% of 6474 prices for timeframe 2hour have been processed. Restarting.
[DEBUG] Training timeframe 2hour, restart count:2
[DEBUG] Coin BTC-GBP: 100% of 6474 prices for timeframe 2hour have been processed. Saving training status then exiting.


[DEBUG] Training timeframe 1hour, restart count:0
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 4hour, restart count:1
[DEBUG] Coin BTC-GBP: 25.0% of 3150 prices for timeframe 4hour have been processed. Restarting.
[DEBUG] Training timeframe 4hour, restart count:2
[DEBUG] Coin BTC-GBP: 100% of 3150 prices for timeframe 4hour have been processed. Saving training status then exiting.


[DEBUG] Training timeframe 1hour, restart count:0
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 8hour, restart count:1
[DEBUG] Coin BTC-GBP: 25.0% of 1619 prices for timeframe 8hour have been processed. Restarting.
[DEBUG] Training timeframe 8hour, restart count:2
[DEBUG] Coin BTC-GBP: 100% of 1619 prices for timeframe 8hour have been processed. Saving training status then exiting.


[DEBUG] Training timeframe 1hour, restart count:0
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 12hour, restart count:1
[DEBUG] Coin BTC-GBP: 25.0% of 1050 prices for timeframe 12hour have been processed. Restarting.
[DEBUG] Training timeframe 12hour, restart count:2
[DEBUG] Coin BTC-GBP: 100% of 1050 prices for timeframe 12hour have been processed. Saving training status then exiting.


[DEBUG] Training timeframe 1hour, restart count:0
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 1day, restart count:1
[DEBUG] Coin BTC-GBP: 25.0% of 1049 prices for timeframe 1day have been processed. Restarting.
[DEBUG] Training timeframe 1day, restart count:2
[DEBUG] Coin BTC-GBP: 100% of 1049 prices for timeframe 1day have been processed. Saving training status then exiting.


[DEBUG] Training timeframe 1hour, restart count:0
[DEBUG] Coin BTC-GBP: 25.0% of 13121 prices for timeframe 1hour have been processed. Restarting.
[DEBUG] Training timeframe 1week, restart count:1
[DEBUG] Coin BTC-GBP: 24.8% of 149 prices for timeframe 1week have been processed. Restarting.
[DEBUG] Training timeframe 1week, restart count:2
[DEBUG] Coin BTC-GBP: 100% of 149 prices for timeframe 1week have been processed. Saving training status then exiting.
```
16 changes: 16 additions & 0 deletions pt_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -272,6 +272,8 @@ def restart_program():
except Exception:
pass

# Enabling DEBUG_PTTRAINER will print a couple of extra lines of debug during training to illustrate a potential issue.
DEBUG_PT_TRAINER = True

the_big_index = 0
while True:
Expand Down Expand Up @@ -371,6 +373,10 @@ def restart_program():
else:
timeframe = tf_list[2]#droplet setting (create list for all timeframes)
timeframe_minutes = minutes_list[2]#droplet setting (create list for all timeframe_minutes)

# Print out the timeframe being selected for training
print(f"[DEBUG] Coin {coin_choice}: Training timeframe {timeframe}, restart count: {restarted_yet}")

start_time = int(time.time())
restarting = 'no'
success_rate = 85
Expand Down Expand Up @@ -1269,6 +1275,12 @@ def restart_program():
which_candle_of_the_prediction_index += 1
try:
if len(price_list2)>=int(len(price_list)*0.25) and restarted_yet < 2:

# Print debug to indicate % of prices processed
# Given the if statement above, this will trigger three times
debug_pct_complete = round((price_list_index / len(price_list)) * 100.0, 1)
print(f"[DEBUG] Coin {coin_choice}: {debug_pct_complete}% of {len(price_list)} prices for timeframe {timeframe} have been processed. Restarting.")

restarted_yet += 1
restarting = 'yes'
break
Expand All @@ -1277,6 +1289,10 @@ def restart_program():
except:
restarting = 'no'
if len(price_list2) == len(price_list):

# Print debug
print(f"[DEBUG] Coin {coin_choice}: 100% of {len(price_list)} prices for timeframe {timeframe} have been processed. Saving training status then exiting.")

the_big_index += 1
restarted_yet = 0
print('restarting')
Expand Down