New @ Stockfuse – Transaction Cost Model

Most virtual trading games execute orders at the last observed market price. This can be problematic if the security being traded is illiquid or the size of the order is large. As many readers may know, stock trading involves a bid/ask spread – investors buy at ask prices and sell at bid prices. Of course, the ask price is higher than the bid price, allowing market makers to capture a profit. On many simulated trading game platforms, however, it is possible for investors to buy at bids and sell shortly afterwards at the slightly higher asks, producing risk-free profits. Stockfuse mitigates this issue by incorporating bid/ask into our execution prices, but we can do a lot better.

It is well known that even large cap funds lag their paper portfolios by 0.5% a year on average, because of a myriad of implicit transaction costs. At Stockfuse, we are committed to providing accurate and realistic trading performance that’s reproducible in the real market. To that end, we are introducing a suite of proprietary Transaction Cost Models (also called implementation shortfall models). These new models enable us to calculate the “market impact” for each order, allowing us to provide far more realistic execution prices. To our knowledge, such advanced models have never been used by simulated trading game platforms before.

The inner workings of the model is beyond the scope of this blog post, but intuitively:

  • Stocks with larger bid/ask spreads are more costly to trade;
  • If you buy or sell a large quantity of a stock at once, the execution cost tends to be higher;
  • If a stock is illiquid (i.e., trading volume in the market is low), the trade will be expensive;
  • When volatility is high, execution costs tends to be higher as well.

For most users, this new model will not make much difference, if any. However, if you execute 100,000 shares of a penny stock, you’ll see the execution price deviating from the last observed market price, just like in real life. As a simple example, buying 100 shares of Morgan Stanley can be easily absorbed by the market, but buying 50,000 shares of MS at a rapid pace will result in some temporary supply/demand imbalance and raise the execution price by about 0.08%.

We expect these new models to completely remove unrealistic liquidity arbitrage trades on Stockfuse. Your P&L will also be much closer to what you’ll actually be able to accomplish in the real market, providing you with more confidence in your strategies.

Appendix: The Stockfuse Transaction Cost Model accounts for the following components:

  • Instantaneous Impact: The cost associated with spread crossing. This is approximately half the bid-ask spread if you always buy at asks and sell at bids; however, we account for the fact that most investors can execute at better levels than bids and asks.
  • Transient Impact: This reflects the temporary imbalances between supply and demand caused by the trade. Intuitively, a buy order raises a stock’s price temporarily in order to attract sellers.
  • Permanent Impact: This is the final change in the equilibrium price caused by the trade. Intuitively, if you buy a stock, other market participants may perceive your action to indicate that you believe the fair value of the stock should be higher. As a result, the equilibrium price, after the trade has completed, will not return to the original level, but should stay slightly higher.

Stockfuse provides Transaction Cost Models for different trading styles, such as VWAP and With Volume. We also allow customization of trading horizon (for VWAP) and participation rate (for With Volume).