Hi Clestian,

I think you can do that with max(). You might need to use min() for short positions.

# initial set trailing stop for long position  
context.longstop = data[stock].price * (1 - .05) # 5% stop-loss price

# update: increase (never decrease) stop-loss  
context.longstop = max( context.longstop, data[stock].price * (1 - .05) ) # 5% target

# test stop-loss  
if data[stock].price < context.longstop:  
   # sell everything  

That didn't work for me... putting both context statements in my initial data (not sure where to put the update statement) seemed to cause an error with the def handle_data section of the script. However, I think I'm getting the desired result with the backtest below... slightly different buy/sell criteria... and the short stop is basically useless at this point;

next thing to do is implement shorting...

So, the initial goal was to follow the price/share and have a trailing stop loss follow the price... What I'm doing here is using the opening price multiplied by my trailing stop loss value (.99 long, 1.01 short) to establish a stop loss. If the price goes below (long) or above (short) the stop loss, the position sells. Trading SPY vs. the benchmark... not too bad a margin (70.5% vs 46.5%) for trading the same index against each other...

any suggestions for improving my buy/sell criteria?

I'm going to attempt allowing the screen to run on multiple companies and see what happens... I'll probably be posting it to another discussion.

You could try a breakout signal. For example, if it's a 30-day high, use that as a buy/long signal, and if it's a 30-day low, use that as a sell/short signal. You can do this by storing the last 30 highs and lows and checking if the current price exceeds the high or falls below the low.

So this example doesn't work great, it's not representative of the strategy, just something you can play around with. I just slightly modified your example, you can see a few new lines added and some adjusted if statements.

Also, you could try using stop or limit orders instead of checking every day to see if your price goes out of certain bounds. Although, you will probably have to update the stop or limit orders, but just something to think about. See here: https://www.quantopian.com/help#ide-ordering