Is price near fix to cover all price ranges

[movy]

The current is_price_near implementation still gives false positives for mid-range priced assets, e.g. ~0.0250, which once again leads to totally unrealistic backtests:

I have updated the method to be more universal + added test cases.

The following code explains the logic behind my formula:

def is_price_near_old(order_price, price_to_compare, threshold_ratio=0.0001):
    """Check if the given order price is near the specified price."""
    percentage_threshold = threshold_ratio * price_to_compare
    # Enforcing a minimum threshold for smaller numbers
    fixed_threshold = 0.001
    threshold = max(percentage_threshold, fixed_threshold)
    return abs(order_price - price_to_compare) < threshold


def is_price_near(order_price, price_to_compare, percentage_threshold=0.0001):
    """
    Check if the given order price is near the specified price.
    Default percentage_threshold is 0.01% (0.0001)
    We calculate percentage difference between the two prices rounded to 4 decimal places, 
    so low-priced orders can be properly compared within 0.01% range.
    """
    return round(abs(1 - (order_price / price_to_compare)), 4) <= percentage_threshold

for prices in [[60000, 60000], [30000, 30005], [30000, 29800], [200, 200.01], [10000, 10001], [0.0250, 0.0249], [0.007386, 0.007385]]: 
    order_price = prices[0]
    price_to_compare = prices[1]
    print(order_price, 'vs', price_to_compare, round(abs(1 - (order_price / price_to_compare)) * 100, 4), '% difference')
    print('is_price_near_old:', is_price_near_old(order_price, price_to_compare))
    print('but is it really within 0.01%?', is_price_near(order_price, price_to_compare))
    print('-----')

Output:

backtest:~/jesse-bot git:(main) ✗ python price_near_test.py
60000 vs 60000 0.0 % difference
is_price_near_old: True
but is it really within 0.01%? True
-----
30000 vs 30005 0.0167 % difference
is_price_near_old: False
but is it really within 0.01%? False
-----
30000 vs 29800 0.6711 % difference
is_price_near_old: False
but is it really within 0.01%? False
-----
200 vs 200.01 0.005 % difference
is_price_near_old: True
but is it really within 0.01%? True
-----
10000 vs 10001 0.01 % difference
is_price_near_old: True
but is it really within 0.01%? True
-----
0.025 vs 0.0249 0.4016 % difference
is_price_near_old: True  <---- current fn gives true
but is it really within 0.01%? False  <----- but should be false
-----
0.007386 vs 0.007385 0.0135 % difference
is_price_near_old: True
but is it really within 0.01%? True

Backtest with the updated method:

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[saleh-mir]

I have modified this function so many times to improve it. It seems like it's not perfect yet. I'm going to try yours with my unit tests. So I'm keeping this open. I'm sorry about the delay.

[saleh-mir]

So I digged into your code and compared it to the existing one. It is certainly better. However, no matter how much I tried I couldn't get it to pass for this assertion too:

# for meme coins such as PEPE
assert jh.is_price_near(0.00000709, 0.00000708) == True

Any ideas?

[saleh-mir]

Or maybe we shouldn't expect this value to be true at all?