Trade Population Weights

Table of Contents

A synthetic trade population is realistic when its mix of products resembles a real book. This note records a weighted distribution across the ORE product set, so a generator can size each product rather than sample uniformly.

1. Summary

ORE's product catalogue defines 88 products, of which 86 are distinct instruments. A population that samples them uniformly is not a book: it holds as many rainbow options as vanilla swaps, which no institution's book resembles.

The distribution below assigns a weight to every one of the 86, grouped by asset class. Interest rate products carry 42% of the trades, foreign exchange 25%, equity 12%, credit 6%, commodity 5%, bonds 4%, and inflation and scripted exotics 3% each.

Two rules shape it. The weights follow trade count rather than notional, because a population is sized in trades. Every product carries a non-zero weight, so a generated population exercises every pricing path at least once.

The weights are a modelling assumption, not a measurement. They are stated so they can be argued with and replaced.

2. Detail

2.1. Why notional cannot set these weights

The public sources report notional outstanding by asset class, and notional is the wrong measure here.

A single interest rate swap carries a notional that hundreds of FX forwards do not reach. Weighting by notional would therefore give a population far too few FX trades and far too many interest rate trades, because it measures the size of exposures rather than the number of contracts. The Bank for International Settlements puts interest rate derivatives at 79% of notional outstanding; that figure is correct and it does not answer the question a population asks.

The weights below are trade counts. They are informed by the notional shares, by which products a dealing institution transacts most often, and by judgement. No public source publishes a trade-count distribution at this granularity.

2.2. Why no product carries zero

A weight of zero removes a product from the population, and with it the only exercise its pricing path gets.

The smallest weights below are therefore floors rather than estimates. A weight of 0.01% is a statement that the product is rare and must still appear. A generator honouring these weights should round up to at least one instance of every product, whatever the population size, so that a tiny preset still covers the catalogue.

This is the one respect in which the distribution is deliberately not realistic. A real book of a thousand trades holds no ascots at all.

2.3. Why a weight for Swap is not yet an instruction

The weights are stated at the granularity of ORE's trade types, and for most products that is the granularity a desk uses. For a few it is not, and those few carry the largest weights.

A desk treats a Bermudan swaption, a cross-currency swap, a tenor basis swap and a vanilla fixed-float swap as four distinct products. It quotes them separately and reports them on separate lines. ORE prices all four through Swap and Swaption, because to a pricing library they are one builder with different legs and exercise data. Both views are correct for their own purpose.

The consequence is that the 19% on Swap is not directly actionable. It does not say what proportion is OIS, cross-currency or tenor basis, and that choice decides which curves and discounting paths a generated population exercises. The same applies to Swaption, where exercise style selects a different pricing model entirely, and to a lesser degree to CapFloor and Bond.

Closing this needs a desk-facing product classification that ORE Studio does not have yet. Until it exists, treat the weight on a trade type covering several desk-facing products as a total to be divided, not as a single population to generate.

2.4. The distribution

2.4.1. Interest rate — 42%

Product Weight Role in the population
Swap 19.00% Vanilla fixed-float and OIS. The largest single population and the curve backbone.
Swaption 4.00% European exercise. Anchors the volatility surface calibration.
Callable Swap 3.50% Bermudan exercise embedded in a swap. The primary rates exotic.
Cap/Floor 3.00% Ibor caps and floors. Anchors the cap volatility surface.
Zero Coupon Swap 2.50% Single exchange at maturity.
Forward Rate Agreement 2.50% Short-tenor curve anchor.
Flexi Swap 2.00% Variable notional schedules.
Balance Guaranteed Swap 1.50% Notional tracking an amortising pool.
Rate Digital Option 1.50% Digital rate payoff.
Cash Position 1.50% Cash balance and money market leg.
Knock Out Swap 1.00% Barrier extinguishing a swap.

2.4.2. Foreign exchange — 25%

Product Weight Role in the population
FX Forward 7.00% Outright linear exposure. The highest trade count of any FX product.
FX Swap 6.50% Near and far leg. Liquidity and curve management.
FX Option 4.00% Vanilla European and American. The volatility surface anchor.
FX Barrier Option 1.60% Single barrier knock-in and knock-out.
FX European Barrier Option 1.20% Barrier observed at expiry only.
FX Digital Option 1.00% Binary payoff.
FX Touch Option 0.90% One-touch and no-touch.
FX Asian Option 0.80% Average rate. Path dependent.
FX Double Barrier Option 0.60% Two barriers bounding a range.
FX Double Touch Option 0.50% Double one-touch and no-touch.
FX Digital Barrier Option 0.40% Digital payoff conditional on a barrier.
FX KIKO Barrier Option 0.30% Knock-in and knock-out combined.
FX Variance Swap 0.20% Realised variance exposure.

2.4.3. Equity — 12%

Product Weight Role in the population
Equity Swap 2.80% Total return swaps and synthetic prime brokerage.
Equity Option 2.20% Vanilla European and American.
Equity Forward 1.40% Linear forward exposure.
Equity Position 1.20% Cash equity holding.
Equity Barrier Option 0.80% Single barrier equity exotic.
Equity Futures Option 0.60% Option on an equity future.
Equity Asian Option 0.50% Average price.
Equity Digital Option 0.50% Binary equity payoff.
Equity Variance Swap 0.40% Realised variance exposure.
Equity European Barrier Option 0.40% Barrier observed at expiry.
Equity Cliquet Option 0.30% Forward-starting ratchet.
Equity Touch Option 0.30% Touch payoff.
Equity Option Position 0.20% Listed option inventory.
Equity Double Barrier Option 0.20% Two-barrier equity exotic.
Equity Double Touch Option 0.10% Double touch payoff.
Equity Auto Delta Hedged Option 0.10% Option with an embedded delta hedge.

2.4.4. Credit — 6%

Product Weight Role in the population
Credit Default Swap 2.60% Single name. Issuer credit risk and CVA.
Index Credit Default Swap 1.60% CDX and iTraxx. The liquidity anchor.
Index Credit Default Swap Option 0.70% Credit spread volatility.
Credit Linked Swap 0.50% Coupon contingent on a credit event.
Synthetic CDO 0.35% Tranche correlation exposure.
Risk Participation Agreement 0.25% Risk participation agreement.

2.4.5. Bond and fixed income — 4%

Product Weight Role in the population
Bond 1.20% Government and corporate inventory, and collateral.
Bond Repo 0.70% Secured funding and the cash leg.
Bond Position 0.50% Bond holding.
Bond Option 0.40% Option on a bond.
Bond Total Return Swap 0.30% Total return on an underlying bond.
Convertible Bond 0.25% Embedded equity conversion.
Bond Future 0.20% Listed bond future.
Callable Bond 0.15% Issuer call schedule.
Forward Bond 0.15% Forward-settling bond.
Ascot 0.05% Asset swapped convertible option.
Bond Option (Reference Data) 0.05% Reference data variant.
Bond Forward (Reference Data) 0.05% Reference data variant.

2.4.6. Commodity — 5%

Product Weight Role in the population
Commodity Swap 1.40% Fixed against floating.
Commodity Forward 1.30% Linear curve building.
Commodity Option 0.90% European and American.
Commodity Average Price Option 0.60% Average price option.
Commodity Option Strip 0.35% A strip of options across delivery periods.
Commodity Swaption 0.25% Option on a commodity swap.
Commodity Position 0.15% Physical or futures holding.
Commodity Variance Swap 0.05% Realised variance exposure.

2.4.7. Inflation — 3%

Product Weight Role in the population
CPI Swap 1.80% Zero coupon inflation. Curve calibration.
Year-on-Year Inflation Swap 1.20% Year-on-year, carrying seasonality.

2.4.8. Scripted and cross-asset exotics — 3%

Product Weight Role in the population
Generic Scripted Products 0.58% The scripting engine catch-all.
Autocallable Type 01 0.50% Retail autocallable and phoenix structures.
Accumulator 0.35% Structured accumulation.
Target Redemption Forward 0.30% Target redemption forward.
Basket Option 0.25% Multi-underlying basket.
Composite Trade 0.20% A trade composed of other trades.
Total Return Swap 0.20% Generic total return.
Worst Of Basket Swap 0.15% Worst-of basket payoff.
Extended Accumulator 0.10% Accumulator with additional features.
Window Barrier Option 0.10% Barrier active over a window.
Rainbow Options 0.08% Multi-asset best-of and worst-of.
Double Digital Option 0.07% Two-condition digital.
Generic Barrier Option 0.05% Generic barrier framework.
European Option Contingent on a Barrier 0.02% Contingent European payoff.
Exotic Variance and Volatility Derivatives 0.02% Variance and volatility family.
Best Entry Option 0.01% Best entry level over a period.
Performance Option Type 01 0.01% Performance-linked payoff.
Strike Resettable Option 0.01% Strike reset feature.

3. Provenance and status

The product list is derived from ORE's own catalogue, the tradedata inputs of Docs/UserGuide/products.tex, which is the definition of what ORE prices. Every row links to this corpus's page for that product.

The weights are this project's assumption. They were informed by an analysis of a dealer's product mix, adjusted in two respects: entries that are not ORE trade types were removed, and the remainder were re-based onto the catalogue's actual contents. Several products named in that analysis, among them Bermudan swaptions, cross-currency swaps and inflation swaps, are not separate ORE trade types. They are configurations of Swap and Swaption, or leg types within them, so they carry no weight of their own here.

The distribution should be replaced by measurement or by a desk's stated mix when either is available. Until then it is a starting point that is explicit about being one.

4. See also

Emacs 29.3 (Org mode 9.6.15)