Kurt Sakaeda built his trading career around a core that most systematic traders overlook: seasonal patterns are not just a curiosity in agricultural markets. They show up in retail stocks, technology companies, cruise liners, and commodities, and they come with a built-in exit that most discretionary setups do not provide.
Kurt has won the World Cup Trading Championships multiple times, with his first win in 2000. In episode 94 of Better System Trader, he covers how he builds and validates seasonal models, how he avoids catastrophic counterparty risk (including first-hand experience of two broker failures), and the story of how he accidentally cornered a futures market.
Watch the full episode below, then read on for the complete breakdown.
How seasonal patterns work beyond grain markets
Seasonal trading has an obvious logic in agricultural futures: crops can only be planted and harvested at specific times of year, so supply-demand dynamics follow a predictable annual cycle. What surprised Kurt early in his career was how far beyond agriculture the patterns extended.
Retail stocks peak in early December. The anticipation of Christmas sales front-runs the actual quarter-four earnings, so investors bid retailers up heading into the season and rotate out after. Hershey’s Candy has a reliable pattern into Valentine’s Day. Technology stocks face pressure around US tax day because employees who have received shares and options need cash to pay taxes, and they sell shares to fund the payment. Cruise liners get booked in January when northern hemisphere weather is cold and people are thinking about vacations.
The common thread is causation. When Kurt spots a seasonal pattern, he looks for the underlying reason it should persist. If the reason is structural, tied to human behaviour, tax calendars, harvest cycles, or physical demand drivers, he treats the pattern as tradeable. If it is just a historical coincidence with no mechanism, he ignores it.
This causation requirement is also his answer to the curve-fitting problem. He cannot rule out that he has been occasionally lucky. But seasonality is not a wave fitted to historical data without explanation. It is a forward-looking model based on repeatable real-world conditions. Grain is expensive in summer because supply is low and demand is high. It gets cheaper in the fall because the harvest arrives. You are not blindly betting on a pattern. You are betting on supply and demand.
How Kurt builds a seasonal model
The process is straightforward. Download the full price history of the instrument, average the returns by business day across all years, fill in holes for holidays, and look for a consistent wave pattern. When a wave emerges, the model generates an entry at the seasonal high and an exit at the seasonal low, or vice versa.
For stocks, he adds two filters to limit damage. He will not go long a company with no earnings and no assets, regardless of how good the seasonal pattern looks. His reasoning: he will need to hold the position for months to capture the full cycle, and a company with no earnings could disappear before the exit arrives. His Enron example: the seasonal system did not flag Enron, but his options system picked up a large move and triggered. He shorted calls. Had he been paying closer attention to the earnings data, he would have known the company had no real earnings, which would have kept him out entirely.
On the short side, the logic is similar but inverted. If a company is generating strong earnings and growing, he does not want to be short it, because a highly profitable company can sustain a rally well beyond what the seasonal pattern predicts. He follows this rule except for two notable cases: Bear Stearns and Lehman Brothers in 2008, where a broker’s advice to hold positions for tax purposes caused him to stay in beyond his planned exit. Both went to near zero. The filter works until it doesn’t, and discipline to stick with the model matters more than the model itself.
Combining seasonal models with disaster trades
Kurt runs two parallel approaches. The core is the seasonal model. The second is what he calls the Time Magazine system: looking for situations where a stock or commodity is in the headlines for a disaster and using the overreaction as a contrarian entry.
When Apple was on the cover of a magazine following Steve Jobs’ death and the stock was at a low, Kurt saw the set up. The company bought it back and it went up over $50 per share within a couple of months. When Volkswagen dropped from around $38 to $24 following the diesel emissions scandal, Kurt calculated that the non-diesel business alone was worth more than $24 a share, and the stock was paying $4 in dividends. He bought it and held it until January.
His preference is when the disaster trade and the seasonal model agree. If a grain market is in the news for drought damage right at the point in the seasonal cycle where prices typically peak before falling into harvest, both approaches point in the same direction. The seasonal model provides the exit. The disaster entry provides a timing anchor.
He has a specific dislike for non-seasonal disaster trades because without a seasonal model to provide an exit date, he has to spend evenings running chart analysis and looking at the position from every angle to figure out when to leave. The work is much heavier. With a seasonal model, time and the natural weight of the supply-demand cycle does most of the work for him.
The counterparty risk problem: Man Financial and Peregrine
Kurt was burned twice by broker failures. Man Financial collapsed. Peregrine Financial went under. In both cases, he had money at the failed firm, and in both cases he lost it. But the damage was manageable because his capital was spread across multiple institutions.
A friend named Ed had all his money at Man Financial. While things were being resolved, Ed had to borrow money from his parents to pay rent. Another friend, Bob, had followed Kurt’s earlier advice to split money across three brokers. When Peregrine failed the following month, Bob lost one-third of his capital. He blamed Kurt and said: if you had told me to use four, I would have only lost 25 percent.
Kurt’s counterparty monitoring approach: if the broker or bank holding your money is a publicly traded company, you can check its stock price. A famous trader he knows spotted that Man Financial’s price chart looked like it was heading toward zero. She closed her account within months before the collapse. You don’t need insider information. The market’s own assessment of the company’s stock price tells you something meaningful about confidence in its financial health.
His practical framework: spread capital across at least five or six institutions, each with a different purpose. Some for trading, some as savings, some designated for future capital expenditures. If one fails, the others remain intact.
How Kurt accidentally cornered a futures market
Early in his trading career, Kurt was using a seasonal model on lumber futures contracts six to seven months out, which had very low open interest. The model signalled a position. He took it. The next morning the model said to add. He added. Over several days he built up to five contracts.
The morning after reaching five contracts, he received phone calls from compliance departments at the introducing broker, the clearing broker, and regulatory staff. They informed him he had cornered the market. He owned five of eight open contracts: 62.5 percent. Technically, that meets the definition.
The episode taught him a practical lesson about liquidity and open interest. Thin futures markets in distant months can have very low open interest, and a systematically generated position can inadvertently become a significant fraction of the total market. He has been more careful about checking open interest relative to his intended position size ever since.
Scaling into new ideas and managing risk over time
Kurt’s risk approach has become more conservative as he has accumulated capital. When he started, he put everything he had into McDonnell Douglas stock after an airliner crash. It worked, but he acknowledges it was the full extent of his tuition money and it took him three years to find another winning trade after that initial success.
Now, when he develops a new idea, he trades one lot to see how it handles in live conditions. Then two lots. Then three. He uses correlation coefficients across his futures portfolio to measure how positions relate to each other and whether adding a new one genuinely reduces risk or adds to an existing exposure. Long crude oil and long gasoline is not diversification. Long crude oil and short gasoline has a negative correlation of 0.95 or higher, which means it can carry much larger sizing without proportionally increasing portfolio risk.
His closing principle: brave enough to try something new, but not brave enough to trade it at full size until you understand how it behaves. The work and the execution. Those are the two things he consistently sees traders skipping when they run into trouble.
Related episodes
- Seasonality trading with Jay Kaeppel
- Strategy development with Perry Kaufman
- Algorithmic forecasting with Larry Williams
Get the show notes & transcript
Want to learn how to build robust seasonal strategies and protect your capital from structural risks? Subscribe to the Better System Trader podcast for weekly interviews with the world’s top systematic traders and quantitative researchers.
