Specific values for stop loss

[ironstone05]

Hi, I have just started with algo trading and use this library to backtest my strategies. Its really great. However, I am having trouble with one specific thing.
I am trying to set the stop loss (and profit) to an exact value instead of a percentage. For eg., for the particular scalping strategy I am testing, I want to set the stoploss to the low of the previous candle and take profit at 1.5 times the risk. Since the margins are low, I want to have better control over the stop loss. Is there a way to set it in such a manner? I went through the documentation and think that either the adjust_sl_func_nb or the stop_exit_price functions are to be used maybe. However, I am unsure on the implementation. It would be great if I could get some help.
Currently, this is the code I am using to backtest:
pf = vbt.Portfolio.from_signals(price, entries, sl_stop= 0.0001, tp_stop= 0.0002, init_cash=10000)

Thanks a lot and kudos on this wonderful implementation! The feature set is so very deep and I am sure I have a long way to go before I understand everything about the library.

[polakowo]

Hi @ironstone05, most arguments can be specified as arrays. Both sl_stop and tp_stop are percentages of the stop entry price. The default stop entry price is close, meaning even though you ordered at the current opening or any other price, the threshold will be computed from the current close (because if the stop is being hit in the same candle, vectorbt wouldn't know whether it's after the entry - we have no intra-candle data - so we use close just for safety). You can change that with stop_entry_price (see enum https://vectorbt.dev/api/portfolio/enums/#vectorbt.portfolio.enums.StopEntryPrice), also globally in vbt.settings['portfolio']['stop_entry_price']. Now, you need to bring both stop values relative to the entry price. For sl_stop it would be

sl_stop = (low.vbt.fshift(1) - entry_price) / entry_price
sl_stop[sl_stop > 0] = np.nan  # previous low is already higher than now? disable stop loss
sl_stop = sl_stop.abs()  # only positive values are allowed

[ironstone05]

Thank you so much! Really appreciate the quick response!

[ironstone05]

I am assuming fshift here is the same as shift in pandas, so implementing it that way should not be an issue.
This is how I have it implemented currently and I think it means the same thing:

sl_stop= abs(price_low.shift(periods=1) - price) / price
where 'price' refers to the Pandas series with the prices I am entering with
and 'price_low' refers to the Pandas series with the low prices

Do let me know if my understanding is wrong. Thanks!

[polakowo]

@ironstone05 this may work, but ensure that price_low is not higher than price (and you probably want stop_entry_price='price'), otherwise, any positive or negative difference between both (such as -0.25 and 0.25) would mean the same thing.

[cryptoliciousdev]

how do i get sl_stop and tp_stop to work with/in this code? like even just a simple sl_stop = 0.05 and tp_stop = 0.10 ?

pf_kwargs = dict(size=np.inf, fees=0.0008, freq='1h')
pf = vbt.Portfolio.from_signals( df_oneCoin['close'],my_long ,my_short , short_entries = my_short , short_exits = my_long , init_cash=100,sl_stop=0.15,sl_trail=0.6, **pf_kwargs)

print('total profit')
print(pf.total_profit())
pf.stats();

i see parts come up in the discussion but never one complete example to reference anywhere....thanks in advance