How do you prevent wash sales in your algorithm?

Hi Cole,

I most be totally lost but it's surprising to me to read that an algo that does intraday trading would ban trading an asset for 30 days if it ever have a losing trade with it during the day.

I'm missing something.

Fred

This might be helpful:

https://www.tradelogsoftware.com/resources/wash-sales/

Ok so the 30 day thing is an IRS rule. Now I understand why it makes sense. 😊

So, yeah, it is very easy to implement… And you can still use wash sales to some extent to “balance” realized capital gains and losses tax-wise.

Fred

Fred Painchaud I also would like to implement a wash sales. You mentioned it was easy, what would be the best way to do it?

Hi Brandon,

The best way IMHO to monitor orders is to use OnOrderEvent(). It is an event that gets called for all order events happening in the algo - for all traded assets, using a Universe or not.

My understanding of wash sales is that if you make a capital loss on a trade, you leave that traded asset aside for 30 days. That's it. My understanding may be incorrect. If so, the below will be incorrect.

So if I am correct, the gist of the algo would use OnOrderEvent to spot loosing trades.

def OnOrderEvent(self, orderEvent):
    order = self.Transactions.GetOrderById(orderEvent.OrderId)
    if orderEvent.Status == OrderStatus.Filled: 
        # Use orderEvent.Direction and orderEvent.FillPrice (etc) to track entering and exiting into positions and calculate capital losses
        # Use orderEvent.Symbol to know which asset it applies to
        # The goal is to track the fill price of your positions' entry and then see if you have lost capital when you exit those positions
        # There is most likely also another way to do it with the Portfolio and its SecurityHolding objects, which have the .Profit, .NetProfit and .LastTradeProfit properties. But it would mean polling the Portfolio. There is certainly yet another way to use both, i.e., use OnOrderEvent to track position exits and then use the Portfolio's corresponding SecurityHolding object (using the Symbol object to get it) to get an eventual capital loss without having to compute it.
        # Of course, you have a dictionary of Symbol --> datetime, tracking when a symbol can be traded next. If a given symbol is not in the dict, it can be traded whenever. If it is, then you cannot trade it before the corresponding datetime. That datetime has been set by the logic above to detect capital losses. Once a datetime for a particular symbol is passed, you do not have to care about it anymore since you can then trade it unless you set the datetime again to now + 30 days.

Fred