Research FAQs

Is fund positioning tradeable? Does fund positioning always revert to the mean?

Hedge funds can stay long or short for long periods of time. After all, extended positioning is a sign that lots of traders believe in a trade. We often say, “Crowded trades are like crowded bars … they’re crowded for a reason.”

It’s important to watch for some fundamental or technical catalyst - e.g., weather problems or macroeconomic drivers - that could push prices against fund positioning and force hedge funds to liquidate big positions. This is like someone pulling the fire alarm in a crowded bar. These big speculator flow events are called ‘short squeezes’ or ‘long liquidations’.

Why are there 13 markets in Peak’s Non-Commercial trader grid but 18 in Peak’s Managed Money trader grid?

The CFTC reports data for only 13 markets in its weekly Commodity Index Trader Supplement COT report. This is where we get all data for Non-Commercial category traders.

The CFTC reports data for more markets in its weekly Disaggregated COT report. Additionally, the Intercontinental Exchange (ICE) reports a weekly COT as well, following the same framework as the CFTC for markets like Robusta Coffee, White Sugar, and London Cocoa.

Which are the best commodity markets for trading hedge fund positioning data?

Markets like Wheat, Soybean Meal, and Lean Hogs are exceptional markets for trading hedge fund positioning, i.e., COT reports and Peak’s COT Today analysis send strong and consistently profitable trading signals for these markets. As a Peak Trading Research subscriber, we highlight these trade set-ups for you when they come up.

Peak CTA Ladder Explained (Daily Email)

Peak’s CTA Ladder model provides you with accurate measures of current momentum and trend-following trader positioning (+10/-10 scale), their recent profitability (PNL), and the price levels where these traders will buy or sell additional contracts, driving prices higher or lower.

Momentum and trend-following traders represent a large percentage of front-contract trading volume in agriculture futures. These traders can move markets significantly higher or lower when they have to buy or sell a large number of contracts to follow their systematic momentum strategies.

We call these momentum traders “CTAs” because they often structure their investment companies as Commodity Trading Advisors.

Peak’s CTA Ladder aggregates Peak’s proprietary momentum and trend-following systems to accurately estimate CTA positioning.

Price levels where CTAs will change positions and put upward or downward pressure on markets are detailed in the CTA Flow diagrams in the PDF attached to every CTA Ladder bulletin. Every day we highlight one market’s CTA Flow diagram as an example in the main bulletin.

Our CTA Ladder systems use machine-learning to check their own accuracy and continually evolve. 

Which trading systems does Peak’s CTA Ladder use to model CTA behavior?

Peak’s CTA ladder model uses 20 different momentum and trend-following models. The models include short-term ( < 21-day price signals), medium-term (21 - 50 day), and long-term (> 50 day) price signals. The models are optimized and weighted for each specific market. We find some markets have more long-term profile traders, e.g., Corn and Soybeans, while some have more short-term profile traders, e.g. Bean Oil and Feeder Cattle.

How are Peak’s CTA Ladder scores calculated? Do the scores represent CTA positions?

The CTA Ladder scores represent how long or short the 20 different momentum and trend-following models are for a certain market. If a market is trending upwards (e.g., white sugar in the table above) and all 20 momentum models are long, we give this market a score of +10 (max long). This score allows you to quickly compare momentum trader positions across all agriculture markets.

If momentum CTAs are max long (+10 rating) or max short (-10 rating) does that mean they can’t buy or sell additional contracts?

That is correct, if momentum traders are max long or max short, they have no more “dry powder” to buy and sell futures contracts, they’ve already maxed out their risk budget. They can only continue to ride the trend or exit positions.

Below is an example of a situation in which CTA momentum traders were 100% short Chicago Wheat futures with no risk budget to sell additional Chicago Wheat futures contracts. We highlighted this trading setup for clients and March Chicago Wheat (W H4 futures) rallied +8 cents (+1.4%) during that trading session (ref: CTA Ladder February 9th, 2024).

Are CTAs the same as Non-Commercial or Managed Money traders?

The CFTC’s Non-Commercial and Managed Money categories comprise all hedge funds: momentum funds, macro funds, fundamental funds, relative value funds … the whole hedge fund pie. CTA traders are the most volatile slice of that hedge fund pie, usually accounting for roughly 25% of front-contract traded volume.

But the short answer is yes. Momentum CTA traders are categorized as Non-Commercial and Managed Money traders. They are speculators.

What is the rolling 21-day PNL calculation?

This PNL calculation represents the aggregate modeled trading profits from the 20 momentum trading systems behind our CTA Ladder model for each agriculture futures market over the previous 21 trading sessions (roughly one calendar month).

This PNL calculation can be from long trades, short trades, or a combination of both.

What is RSI?

Relative strength index (RSI) is a popular momentum indicator used in technical analysis that measures the magnitude of recent price changes to evaluate if a market is technically overbought or oversold. The RSI indicator can have a reading from 0 to 100.

Does this CTA ladder work? Does it accurately model CTA behavior?

Yes, this model works very well. You’ll consistently see prices accelerate through the flow zones predicted by our CTA Ladder model. Additionally, the model generates paper profits that match CTA performance indices, i.e., the model does a great job of replicating the actual momentum strategies that CTA traders deploy.

How does Peak’s CTA Ladder model come up with the trading volume (number of contracts) traded at each price level?

There are two different ways that our CTA Ladder model finds these trading volumes:

1.) The model re-calibrates based on observed trading volumes over time. For example, the model will predict a certain number of contracts will be traded at a certain level, and then adjust based on observed volumes traded at that price level as reported by the exchange. The model ‘learns’ and adjusts to become more accurate over time.

2.) The model also does some high-level adjustments based on COT data (for the markets in which COT reports are available). This is more difficult because there is a lot of noise in the COT reports via other speculators (i.e., the Non-Commercial category contains all hedge funds … not only CTA momentum traders). But, that said, Peak’s CTA Ladder model still incorporates COT data to improve through time.

Peak Price Seasonality Explained (Weekly Bulletin)

Peak’s Price Seasonals allow you to anticipate seasonal price trends over the coming weeks and months based on price data from the past 15 years.

Agriculture futures prices have strong and consistent seasonal trends as traders add and remove risk premiums during production cycles throughout the year.

Agriculture traders get nervous during the U.S. springtime ahead of the critical June-July-August production window and are willing to pay more for agriculture futures. These traders will then pay less for agriculture futures later in the summer once final U.S. production prospects are better known.

The same thing then happens again during the South American production cycle; prices rise in the fall months as we again build in a risk premium.

Peak’s forward-looking Seasonal Heat Maps and Normalized Seasonal charts highlight seasonal price trends, basis 15 years of futures prices, so you can confidently anticipate upcoming seasonal market moves.  

How should I interpret the -5 to +5 rating system in Peak’s Seasonal Heat Maps?

A +5 week is a seasonally bullish period when prices tend to rise based on data from the past 15 years. Conversely, a -5 week is a seasonally bearish period when prices tend to fall. This analysis indexes the percentage moves on a -5 to +5 scale for easy comparisons across the different agriculture markets.

I see all the normalized seasonal curves start at 1.00, how is that calculated?

This analysis takes the past 15 years of data and gives every January 1st date a value of 1.00. We then adjust that 1.00 value by the percentage price changes throughout the year for each market and aggregate those 15 lines together into our Normalized Seasonality lines. This shows the seasonal trend throughout the calendar year for each market.

I see some seasonal lines trend up or down all year, how is that possible?

Some markets have strong positive or negative carry when rolled over time (depending on the frequency of calendar curve inversions). We accurately adjust for each market’s carry profile to reflect the real long-term investment returns from holding and rolling a position in those markets. Because seasonal trends can take multiple months to play out (and multiple contract rolls), the carry yield should be considered.

What are the green and red columns at the bottom of the normalized seasonal charts?

These columns highlight particularly strong seasonal periods, i.e., when the 15-year Normalized Seasonal line trends up or down > 0.25% for two or more consecutive days. The green and red columns are consistent with the sharp inclines and declines in the smoothed Normalized Seasonal line that you see in each chart (example below).

Peak Price Patterns Explained (Monthly Email)

Peak’s Price Patterns allow you to anticipate and trade specific historical price patterns, complementary to Peak’s broader Seasonals analysis.

Agriculture futures prices have strong and consistent seasonal trends. There are often ‘high hit rate’ trades within those trends.

For example, above, March Gasoline futures have rallied in 14 of the past 14 years (100%) for the 12 trading sessions starting February 6th. The Wall Street Journal talks about this trade and Peak’s analysis here. See chart below.

How does Peak find these seasonal price patterns?

We've built a proprietary Pattern Search Engine in the Python coding language that looks for 5-70 day patterns with a greater than 85% hit rate across all major Agriculture, Agriculture Spreads, Energy, and Metals futures markets. The Engine analyzes every liquid futures contract, searching for patterns that have repeated consistently for a decade or more.

Peak’s Pattern Search Engine often discovers several price patterns in close proximity to one another (e.g., starting Feb 5th, Feb 6th, Feb 7th … ) or across different contracts. In these situations, it filters for patterns with the highest signal strength. These patterns are the most likely to identify a true seasonal trading edge.

How does Peak calculate those different signal strengths? What makes a signal weak or strong?

To calculate signal strength, Peak’s Pattern Search Engine takes the average price pattern move as a percentage of the series' standard deviation, normalized by the number of days. To say this another way, the Engine assesses if a pattern is strong based on the volatility of the time series and the magnitude of moves that each market typically sees.

Example #1: Sugar moving 3 cents in 21 days is a stronger signal than Soybeans moving 8 cents in 21 days, because of different contract sizes and volatilities between Sugar and Soybeans.

Example #2: Corn moving 10 cents in 5 days is a stronger signal than Corn moving 10 cents in 40 days, because of the different holding periods.

The signal strength calculation summarizes those differences (weak, medium, strong, max).

Do you have a high conviction to trade these patterns? 

Yes, especially if there is a strong price pattern signal in the context of a larger, fundamentally based seasonal move, for example, a ‘Max’ bullish price signal in Soybean Meal during the seasonally bullish month of October.

Conviction is lower for weaker, shorter price patterns that fall during a period with less broad seasonal clarity.

Cross-referencing Peak’s Price Patterns with Peak’s Seasonal Heat Maps and Normalized Seasonal charts is a great starting point.

What's your data source for prices? 

Bloomberg and IQFeed.

What are the trading rules to follow these patterns?

Positions are executed at the market close of the date specified or on the market close of the first open session after for weekends and holidays. Trades are held for the number of open market trading sessions, not calendar days. We’ve found that trading these patterns with a 2:1 profit-stop ratio is the most consistently profitable approach, e.g., risking 10 cents to make 20 cents. This wide berth allows for market volatility and gives the seasonal trend some ‘room to breathe’.