Sometimes you can improve a strategy's performance by simply reducing the portfolio turnover. For example, by entering the top 10% of stocks by alpha signal, but then only exiting when they fall out of the top 20%. You need to be able to monitor your turnover, as the trading costs will be directly related to that. You can do this in Pyfolio, but I like to see the feedback within the backtester.

Here's some code that takes the output of Optimizer, and charts the portfolio's daily turnover. For my shorter term strategies, I have daily turnover in the 20-40% range.

    objective = opt.TargetPortfolioWeights(targets.weight)

    constraints = [  
        opt.MaxGrossLeverage(MaxGrossLeverage),  
        opt.DollarNeutral(MaxNetExposure),  
        ]

    order_ids = algo.order_optimal_portfolio(  
        objective=objective,  
        constraints=constraints,  
        universe=targets.index  
    )  


    # Reporting

    log.info("Today's positions: %s" % (" ".join(str(x) for x in [s.symbol for s in targets.index])))  
    record(Positions=len(targets))

    orders  = pd.Series({get_order(i).sid:get_order(i).amount for i in order_ids.values})  
    prices = data.current(orders.index,'price')  
    buy_order_value  = sum([abs(amount*prices[s]) for s,amount in orders.iteritems() if amount>0])  
    sell_order_value = sum([abs(amount*prices[s]) for s,amount in orders.iteritems() if amount<0])  
    order_value = min(buy_order_value, sell_order_value)  
    turnover = order_value / context.portfolio.portfolio_value  * 100  
    record(Turnover=turnover)