You were right to be suspicious of the results! The ordering method doesn't check for open orders before calculating the number of shares to buy, so you were ending up with a long list of them.

A large order may take several bars to fill, according to the slippage model. Especially an order to buy or short a $1M portfolio! While this order is being filled, handle_data is getting called every minute and your order_target_percent is adjusting its values, creating a queue of open orders. This may also lead to large swings of order amounts.

I fixed this problem by adding a check on line 8 of the code (see the attached backtest). However, if you look at the graph of the cash, you can see that the algorithm shorts $1M - this is only allowed by the broker if you have a margin account (and you'll be paying interest). The line below means you're shorting 100% of your portfolio.

order_target_percent(context.security, -1)  

Do you instead mean to sell the portfolio and close the position if current_price/mav30>1.05? In that case you want,

order_target_percent(context.security, 0)  

In the custom graph, the stock price is getting masked by the cash because its so much smaller. If you want to view the stock price then click on "cash" to temporarily remove it from the graph. To bring it back, click on "cash" again.

Hope that helps!

The intent was to buy when the price/30sma drops below 0.95 and the sell when it rises above 1.05. Then, for good measure short the security as well. I'll start a new backtest with your addition (thank you very much, new to python) and with a smaller starting cash pool. Thanks!

That's still over 300% are we sure nothing else if funky?

I replied to this in another thread, but I had a minute one running that is finished now. I changed line 17 to fix the excessive borrowing. It looks like the strategy performs much better with minute data, it still looks good without the borrowing.