Unfortunately, as documented the actual slippage depends on the volume of the next bar in the future so there would be look-ahead bias I think. But you could code your own rough predictor and use it to break up your order in to small lots. A starting point for an order volume that should have minimim slippage would be to get the volume traded over the last 30 minutes, calculate the average and then order say 5 to 10% of that volume. Repeat every bar until your order is filled. If you get filled at 10% of the volume on the next bar, your slippage in the classic Q model would be 0.1*(0.1)^2 = 0.001 = 0.1%; at 5% it would be 0.1*(0.05)^2=0.025% . (Not sure what Q's new slippage model would do though as the final version isnt on the api help page yet or I have missed it).

@Richard - Thank you. Your post actually goes further - suggests what to do to keep slippage under control.

I decided that, with the FixedSlippage model, I want get_slippage(sid) to return the half-spread (not the spread) for the security sid. Then I can simply add it to the most recent price data[sid].price to obtain an estimate of the net price I'm going to pay. Then the code to calculate and order as many shares of sid as cash allows becomes

cpt, cps=get_commission() # commission per trade, per share  
hs=get_slippage(sid) # half-spread, assuming the FixedSlippage model  
p=data[sid].price  
n=int((cash-cpt)/(p+hs+cps))  
if n>0:  
    order(sid, n)  

It would be nice if get_slippage(sid) returned something that could be added to the current price (or subtracted if I'm selling) to obtain an estimate of the actual transaction price, regardless of the slippage model used, even if user-defined. Can this be done? How?

André, I think Richard is giving excellent advice.

Slippage is, by definition, unpredictable. The slippage model that Quantopian uses by default is a decent estimation, but it is far from perfect. The real world will behave differently every time, and it cannot be predicted. You can't know in advance what other buyers and sellers will be in the market, what news is breaking, etc. Richard is suggesting a way to manage the risk of slippage. Managing the uncertainty of slippage is a viable strategy, but removing the uncertainty is not.

@Dan - We know that the exact price we'll be charged for (each chunk of) our order cannot be known or calculated. But it can be treated as a random variable that follows some distribution, and estimated, ie. its expected value (or maximum, or the 95th percentile) found, can't it?

You seem to argue that the future is entirely unknowable. Do you watch weather forecasts? Do you factor them into your decisions? But the weather is more chaotic, and less predictable, than financial markets, isn't it?