Treasury Yield Curve

Integrate Treasury yield data via FRED API. Yield-adjusted Kelly sizing, opportunity cost filtering, and collateral optimization based on the yield curve shape.

Ultra Feature. Requires an Ultra subscription. Get started at api.mathematicalcompany.com

Treasury Yield Curve

Prediction market positions tie up capital that could earn risk-free yield in Treasuries. When rates are high, a trade needs more edge to justify the opportunity cost. Horizon’s treasury module provides a live yield curve feed (via FRED), yield-adjusted Kelly sizing, breakeven edge calculation, and collateral optimization. All math is in Rust.

Overview

Yield-Adjusted Kelly

hz.yield_adjusted_kelly() reduces Kelly size by the risk-free rate opportunity cost.

Opportunity Cost

hz.opportunity_cost() and hz.breakeven_edge() quantify the hurdle rate for a trade.

Live Yield Feed

TreasuryFeed polls FRED for the full yield curve (1M through 30Y).

Collateral Optimizer

hz.collateral_optimizer() adjusts capital allocation based on curve steepness.

---

Core Functions

All math runs in Rust. Every output is guarded against NaN/Inf.

hz.yield_adjusted_kelly

Standard Kelly fraction minus the annualized opportunity cost of locked capital.

import horizon as hz

fraction = hz.yield_adjusted_kelly(
    prob=0.60,
    price=0.50,
    rfr=0.05,        # 5% risk-free rate
    duration_days=30,
    fraction=0.5,     # half-Kelly
)
print(f"Adjusted Kelly: {fraction:.4f}")

Parameter	Type	Default	Description
prob	float	required	Estimated probability of the event
price	float	required	Current market price
rfr	float	required	Risk-free rate (annualized decimal)
duration_days	float	required	Expected holding period in days
fraction	float	0.5	Kelly fraction (0.5 = half-Kelly)

Returns 0.0 if the edge does not exceed the opportunity cost.

hz.yield_adjusted_kelly_size

Convert yield-adjusted Kelly fraction to a dollar position size.

size = hz.yield_adjusted_kelly_size(
    prob=0.60, price=0.50, rfr=0.05,
    duration_days=30, bankroll=10000.0,
    fraction=0.5, max_size=1000.0,
)
print(f"Position size: ${size:.2f}")

Parameter	Type	Default	Description
prob	float	required	Estimated probability
price	float	required	Market price
rfr	float	required	Risk-free rate
duration_days	float	required	Holding period in days
bankroll	float	required	Total bankroll
fraction	float	0.5	Kelly fraction
max_size	float	inf	Maximum position size cap

hz.opportunity_cost

Dollar cost of locking capital at the risk-free rate.

cost = hz.opportunity_cost(notional=5000.0, rfr=0.05, duration_days=30)
print(f"Opportunity cost: ${cost:.2f}")  # ~$20.55

Parameter	Type	Description
notional	float	Capital locked in the position
rfr	float	Risk-free rate (annualized)
duration_days	float	Holding period in days

hz.breakeven_edge

Minimum edge required to beat the risk-free rate.

be = hz.breakeven_edge(rfr=0.05, duration_days=30)
print(f"Breakeven edge: {be:.4f}")  # ~0.41%

Parameter	Type	Description
rfr	float	Risk-free rate
duration_days	float	Holding period in days

---

Feed

TreasuryFeed

Polls the Federal Reserve Economic Data (FRED) API for yield curve data.

import horizon as hz

feed = hz.TreasuryFeed(
    api_key="your_fred_api_key",  # or set FRED_API_KEY env var
    primary_series="DGS10",
)

Parameter	Type	Default	Description
api_key	str	FRED_API_KEY env var	FRED API key
series_ids	list[str]	9 standard maturities	Treasury series to fetch
primary_series	str	"DGS10"	Series used as main price
interval	float	300.0	Poll interval in seconds

Default series: DGS1MO, DGS3MO, DGS6MO, DGS1, DGS2, DGS5, DGS10, DGS20, DGS30.

FeedSnapshot mapping:

• price = primary series yield (decimal)
• bid = short-term yield (DGS3MO)
• ask = long-term yield (DGS30)
• volume_24h = curve slope in basis points (long - short)

---

Pipeline Functions

hz.yield_adjusted_sizer

Size positions using yield-adjusted Kelly each cycle.

sizer = hz.yield_adjusted_sizer(
    rfr_feed="treasury",
    duration_days=30,
    fraction=0.5,
    bankroll=10000.0,
    max_size=1000.0,
)

hz.run(
    name="yield_aware",
    feeds={"treasury": hz.TreasuryFeed(), "market": hz.PolymarketBook("...")},
    pipeline=[sizer, my_strategy],
)

Parameter	Type	Default	Description
rfr_feed	str	"treasury"	Feed name for the yield data
duration_days	float	30.0	Expected holding period
fraction	float	0.5	Kelly fraction
bankroll	float	10000.0	Total bankroll
max_size	float	1000.0	Maximum position size

Returns each cycle:

{
    "size": 245.0,
    "yield_adjusted_kelly": 0.049,
    "breakeven_edge": 0.004,
    "opportunity_cost": 20.55,
    "rfr": 0.05,
}

hz.opportunity_cost_filter

Skip trades where edge does not justify the opportunity cost.

filter_fn = hz.opportunity_cost_filter(
    rfr_feed="treasury",
    duration_days=30,
    min_edge_multiple=2.0,  # need 2x breakeven edge to trade
)

Parameter	Type	Default	Description
rfr_feed	str	"treasury"	Feed name for yield data
duration_days	float	30.0	Holding period
min_edge_multiple	float	2.0	Required multiple of breakeven edge

hz.collateral_optimizer

Adjusts capital allocation based on yield curve steepness.

optimizer = hz.collateral_optimizer(rfr_feed="treasury", max_utilization=0.8)

Parameter	Type	Default	Description
rfr_feed	str	"treasury"	Feed name
max_utilization	float	0.8	Maximum capital utilization

Steeper curves reduce allocation (higher opportunity cost of capital).

---

Examples

Yield-Aware Market Making

import horizon as hz

def my_mm(ctx):
    if ctx.feed is None or ctx.feed.price <= 0:
        return []
    # Use yield-adjusted size from upstream pipeline
    size = ctx.params.get("size", 10)
    return hz.quotes(fair=ctx.feed.price, spread=0.04, size=size)

hz.run(
    name="yield_mm",
    exchange=hz.Polymarket(),
    feeds={
        "market": hz.PolymarketBook("will-fed-cut-rates"),
        "treasury": hz.TreasuryFeed(),
    },
    pipeline=[
        hz.yield_adjusted_sizer(rfr_feed="treasury", bankroll=10000),
        hz.opportunity_cost_filter(rfr_feed="treasury", min_edge_multiple=2.0),
        my_mm,
    ],
)

---

Mathematical Background

Yield-Adjusted Kelly

Standard Kelly: f = (p - q) / (b - 1)* where p is probability, q = 1-p, and b is the payout odds.

Yield adjustment subtracts the annualized opportunity cost:

f_adj = max(0, f - rfr * duration / 365) * fraction*

When the risk-free rate is high relative to the edge, the adjusted fraction drops to zero, meaning the trade is not worth taking.

Breakeven Edge

The minimum edge that justifies locking capital:

breakeven = rfr * duration / 365

At 5% rates and a 30-day hold, the breakeven is about 0.41%. Any edge below that is better invested in T-bills.

Opportunity Cost

Dollar cost of capital deployment:

cost = notional * rfr * duration / 365

This represents the forgone interest on the locked collateral.