Yeah I've been reading a lot of stuff on option trading for work, since our option transaction fees have grown significantly low due to high turnover volumes as a market maker for a number of instruments. Here're some things from the book that's really caught my attention:
"
' Although it is always perilous to assume that the future will be like the past, it is at least instructive to find out what the past was like. Experience suggests that for predicting future values, historica data appear to be quite useful with respect to standard deviations, reasonably useful for correlations, and virtually useless for expected returns.' - William Sharpe
My view is that the theory is correct in the same way that the theory that the earth is spherical is correct. The earth is not perfectly spheroid, but from a sufficient distance it looks spherical. The planet's bumps and irregularities are small and the idea of a spherical earth is a far better descriptive theory than the one it replaced, the idea that the earth was flat. Using the spherical earth theory has allowed us to do many useful things such as flying around the world. However, a pilot who blindly believed in the theory would put us in grave danger: his theory has no place for 'anomalies' such as mountains.
The efficient market theory is similar.
... a good trader makes money by exploiting the small deviations from the theory without bothering to argue that markets are totally inefficient.
"
Yep, that is pretty much how it is, reality vs. academic theories.
Prerequisite mathematical understanding
To fully understand everything in the book, understanding of basic calculus and probability theory is required. The math involves mostly partial derivatives and basic integrations.
Basic calculus tutorial: Karl's Calculus Tutor
Basic statistics tutorial: StatTrek
The math is needed for theoretical valuations, volatility estimates, and strategy P/L expectations. Important stuff.
Defining an edge
Sinclair does a good job of explaining the concept of trading with a positive expected value.
EV = P(win) * Payoff(win) + P(loss) * Payoff(loss)
where
P(win) = probability of winning
Payoff(win) = expected payoff per winning trade
P(loss) = probability of losing
Payoff(loss) = expected payoff per losing trade
From my experience, really understanding this concept is what makes someone make money over time. Also a really good point Sinclair's made is that while backtesting shows if something's worked in the past, "we need to know why they will do so in the future." The point being, there're an infinite number of combinations of trade ideas, parameters, and products that have produced successful trades in the past; and without knowing WHY something's worked in the first place, it is very tough to know if it's stopped working. This is a very realistic risk not recognized by many in the algorithmic trading field.
Taleb has mentioned that a good trading strategy must hold a robust theory, empirical evidence (backtesting) is a secondary requirement.
Market Making Techniques
Some good concepts are mentioned around inventory (risk) management and quoting widths. Keep in mind that a number of high frequency trading algorithms hold market making as core.
Volatility Trading
While the idea of implied volatility converging with historical vol or "volatility arbitrage" is probably discussed to death, the academic guys have made a lot of inpractical assumptions around transaction costs. It is NOT a riskfree arbitrage.
So while our expected return off this trade is
P/L = Vega (Implied Vol - Historical Vol)
it is true only on average. This book explores reality of vega trading while estimating transaction costs from hedging and path taken by the position. Basically, it takes a lot more than that to exploit these deviations, and it could easily turn into a loss with less-than-optimal hedging frequency and/or undesirable path of the underlying while the position is open.
Over all it is one of the better books I've read around financial trading.
"
' Although it is always perilous to assume that the future will be like the past, it is at least instructive to find out what the past was like. Experience suggests that for predicting future values, historica data appear to be quite useful with respect to standard deviations, reasonably useful for correlations, and virtually useless for expected returns.' - William Sharpe
My view is that the theory is correct in the same way that the theory that the earth is spherical is correct. The earth is not perfectly spheroid, but from a sufficient distance it looks spherical. The planet's bumps and irregularities are small and the idea of a spherical earth is a far better descriptive theory than the one it replaced, the idea that the earth was flat. Using the spherical earth theory has allowed us to do many useful things such as flying around the world. However, a pilot who blindly believed in the theory would put us in grave danger: his theory has no place for 'anomalies' such as mountains.
The efficient market theory is similar.
... a good trader makes money by exploiting the small deviations from the theory without bothering to argue that markets are totally inefficient.
"
Yep, that is pretty much how it is, reality vs. academic theories.
Prerequisite mathematical understanding
To fully understand everything in the book, understanding of basic calculus and probability theory is required. The math involves mostly partial derivatives and basic integrations.
Basic calculus tutorial: Karl's Calculus Tutor
Basic statistics tutorial: StatTrek
The math is needed for theoretical valuations, volatility estimates, and strategy P/L expectations. Important stuff.
Defining an edge
Sinclair does a good job of explaining the concept of trading with a positive expected value.
EV = P(win) * Payoff(win) + P(loss) * Payoff(loss)
where
P(win) = probability of winning
Payoff(win) = expected payoff per winning trade
P(loss) = probability of losing
Payoff(loss) = expected payoff per losing trade
From my experience, really understanding this concept is what makes someone make money over time. Also a really good point Sinclair's made is that while backtesting shows if something's worked in the past, "we need to know why they will do so in the future." The point being, there're an infinite number of combinations of trade ideas, parameters, and products that have produced successful trades in the past; and without knowing WHY something's worked in the first place, it is very tough to know if it's stopped working. This is a very realistic risk not recognized by many in the algorithmic trading field.
Taleb has mentioned that a good trading strategy must hold a robust theory, empirical evidence (backtesting) is a secondary requirement.
Market Making Techniques
Some good concepts are mentioned around inventory (risk) management and quoting widths. Keep in mind that a number of high frequency trading algorithms hold market making as core.
Volatility Trading
While the idea of implied volatility converging with historical vol or "volatility arbitrage" is probably discussed to death, the academic guys have made a lot of inpractical assumptions around transaction costs. It is NOT a riskfree arbitrage.
So while our expected return off this trade is
P/L = Vega (Implied Vol - Historical Vol)
it is true only on average. This book explores reality of vega trading while estimating transaction costs from hedging and path taken by the position. Basically, it takes a lot more than that to exploit these deviations, and it could easily turn into a loss with less-than-optimal hedging frequency and/or undesirable path of the underlying while the position is open.
Over all it is one of the better books I've read around financial trading.
0 Reflections:
Post a Comment