Question

Generate a model of a natural phenomenon, or an engineered product or process, or describe an already existing model and evaluate how well the model represents the situation.

Answer 1

In the field of financial markets, let's consider the costs which are incurred at the time of trading. We call them transaction costs.

The transaction cost depends on several factors and can be modeled on the basis of these:

1) Spread: Different between best bid and best ask price at the time of order execution (or the average of these if the order took longer to execute)

2) Participation rate: In general the higher the participation rate of the order wrt total volume at the time, the higher the impact on the prices, since it costs more to trade with higher urgency/aggressiveness

3) Trade size: Normally considered as % of ADV (Average Daily Volume) so that it is normalized when we compare similar order sizes of less liquid and more liquid stocks. Generally, a higher order will signal more to the market about the intent to trade, thus having an adverse impact on the prices

4) Volatility: This is the std deviation of stock returns over a period of time. In general, the more the volatility, the more we have to pay for the order if other factors remain at similar levels.

We can model the Transaction costs on the basis of above using historical trade data. Due consideration needs to be given to the varying roles of these factors in deciding the transaction cost. The data is finally calibrated to real market situations to make it more realistic. This model can go a long way in pre-trade analysis to estimate the cost of trading before sending out an order to the market. Calibrating it on different market scenarios (trending up/down, volatile/non-volatile, etc) lets us ensure the mode doesn't fall apart when market microstructure changes.

Hope the above realistic example helps.