StopMarketOrder Fill Price

Hi I looked at your code and your orders. Your stop market orders were placed 3 days later at 12:00am every time in the code you pasted. I would look at that first. the order should be placed as soon as the trade is placed.

also wouldn't you want your stopmarketorder entry price to be below your buy price or above your short price, ? you are saying 

For Limit orders, when the limit price is reached, the order will be filled at the limit price. For StopMarketOrder, When the stop price is reached, a stop order becomes a market order. The order will be filled at the market price. Forex has quote bar data with bid/ask price. Therefore, the stop market buy orders are filled with ask price and the stop Market sell orders are filled with the bid price.

You can get the bid/ask price with "self.Securities[self.pair].AskPrice" and "self.Securities[self.pair].BidPrice". 
 

Hello Jing Wu ,

my logic here is that I want to enter the trade at the stop price that higher than the current price when market price is equal to that stop price. In Oanda platform a stop order is used for this. That's why the confusion.
I thought "When the stop price is reached" means that market price is equal to this stop price.
I will use LimitOrder also for this entry logic. I hope there won't be a problem in live trading.

Independent of this use case I still want to understand the fill price of StopMarketOrders. Is it the Close price of the daily bar (if the resolution is daily) ?
I implemented your suggestion to look at bid and ask. You are right that most of the cases the fill price is equal to this ask price. However, here is a couple of examples where I cannot relate the fill price to ask price but it's equal to stop order price.

What is the reasoning here?

---
(Apollos Hill hi, it was not a stop 'loss'. I say if market reaches this price, it's a confirmation for me that it would go even higher. Hence, I would buy only at that stop price. I got that I should use LimitOrder for this, Thank you.)

backtest attached.

If the resolution is daily, the order fills based on the daily bar price. 

For stopMarketOrder, FillPrice = max(StopPrice, marketPrice + slip);

If StopPrice is higher than the current market price + slippage, it will be filled at the stop price.

You can check the fill model of LEAN for StopMarketOrder.

Hi there,

I think I'm experiencing the same issue. My stop market order is placed at 329.85 and is filled on the following day when the market goes through, however, the fill price turns out to be the closing price of the bar: 324.56. 

According to the default fill model, shouldn't the stop price be the fill price since it is higher than the closing price?

Hi Hiếu Ngô,
 For StopMarketOrder, a stop order becomes a market order when the stop price is reached. The order will be filled at the market price. A sell stop market order will trigger when the price is equal to or lower than the one set.
Best,
Varad Kabade

Hi Varad Kabade , I appreciate the answer but this is where my question is getting at. If the stop order triggers at market price, why doesn't it fill at the market price at that time which would be the price of my stop order (or very close to it but assuming 0 slippage)?

Jing Wu has already said this in her answer above me: If StopPrice is higher than the current market price + slippage, it will be filled at the stop price.

In my example my stop market order is placed at 329.85, when new data comes in and the closing price is lower than the closing price (324.56) I am expecting for the stop to be hit intraday and my fill price to be the stop market order price: 329.85

Hi Hiếu Ngô 

Stop filling on bar data is far more complex than it seems 😬. There are a few points to note:

• You don't know the order of the HL in the OHLC; you only know the open and close; the intraday bar movements are unknown. 
• The bar events which set the rhythm of time effectively jump through the whole day, there's no movement of the price during the day. E.g. The price could have gapped down at any moment. Orders are placed and fills processed filled at the end of bars. 

We assume the order is filled pessimistically – the worse scenario given the data available. For more accurate fills please switch to a higher resolution of data (e.g. minute or second) to get more accurate stop fills.

Hi Hiếu Ngô,

I'll add a little something to the above great answers.

The relevant source code for your fill is the following:

case OrderDirection.Sell:
    //-> 1.1 Sell Stop: If Price below setpoint, Sell:
    if (prices.Low < order.StopPrice)
    {
        fill.Status = OrderStatus.Filled;
        // Assuming worse case scenario fill - fill at lowest of the stop & asset price.
        fill.FillPrice = Math.Min(order.StopPrice, prices.Current - slip);
        // assume the order completely filled
        fill.FillQuantity = order.Quantity;
    }
    break;

‘prices’ is a structure containing last bar's data points.

Note that what Jing Wu wrote is slightly wrong - she refers to the Buy Stop Market Order fill model, but you are selling.

So, in English:

1- If last bar's Low is less than the stop price

    a) the status of the order is set to Filled

    b) the fill price (what you are interested in) is the min between the stop price and the current price + slippage. Note: the “current” price is the Close here.

    c) the fill quantity is assumed to be the order quantity (so complete execution is assumed)

So now, going back to Jared's comment on pessimistic fills - if you keep in mind that it is not the max, it is the min - everything comes together and makes sense. You do not want a simulation where you get 329$/stock when in reality you could eventually get 324$/stock…

In the world of simulations, backtesting is really a low fidelity (lo-fi) simulation and in lo-fi simulations, you want to be as conservative as possible to limit false positives, i.e., simulations that yield good results but when the model (strategy here) is applied to reality, worse results are very frequent. Even with great care at being conservative (or pessimistic), lo-fi simulations suffer from a high rate of false positives so you do not want to worsen it.

Fred