Modeling Fixing Risk in Interest Rate Swaps

# EURIBOR Fixing Risk


# EURIBOR Fixing Risk












So this is really about EURIBOR as we are in a RFR framework in other major currencies. I'm interested in understanding the exact EURIBOR fixing risk a swap trader really faces if he manages a large book of swaps. So, here are my thoughts:

Suppose my delta risk is in non-overlapping fwd space, hence I can easily calculate the risk in each fwd fixing. Before 11 am EURIBOR publication time, I have some estimate on the upcoming fixing on my curve setup which may or may not be accurate. I know exactly the fixing dv01, which I interpret as the P&L for a 1 bps movement in the coming fixing. After 11 am the fixing is known and a set of floating cash flows are now fixed (generating realized fixing P&L) and hence only carry some discounting risk afterwards.

If I price a new swap after 11 am, I should again see a new fixing dv01 in the 6m fixing bucket right? (suppose we have a 6m EURIBOR swap). Are there any other hidden fixing risks I should take into account or is it really about upcoming cash flows fixing? Example: After the fixing is published should this also affect all other swap rates in the curve as their first period now also are fixed at a different level than the market may have estimated before? Does this impact my delta risk strip after publication?

Happy to discuss/clarify

## Answer by Attack68 (score 2, accepted)

https://quant.stackexchange.com/a/85231

The general principle here is that any IRS with an unpublished fixing will have financial exposure to that publication and fixing value. The PnL event is classed as being the difference between the published value and the market forecast value.

As a practical example, in October 2022, STIBOR 3M was revised, and its (initial) calculation methodology led to large volatility (sometimes 20bps from market forecast). If you traded a 1Y SEK IRS (Annual-Quarterly) before 11am this price would based on a forecast 3M-STIBOR, lets say you paid 1bn 1Y with delta 100k SEK. If the 3M-STIBOR published 20bps higher than expectation the 1Y price in the market would move higher by about 5bps (20 * 0.25) immediately and you would gain roughly 500k SEK on your trade. The forward rates would be unchanged, so this PnL would be attributable purely to the move in the fixing.

What you "see" in your risk is entirely dependent upon how your risk model is configured, and what you want it to do. Different representations are possible, in my opinion some are more useful than others.

> Can you elaborate on the more useful ones? Suppose I trade a swap after 11.am. This should not be sensitive against the already published fixing (only discounting risk)

If you traded a (spot) IRS after 11am, with known fixing, this transaction is well defined, e.g:

```
from rateslib import *  # Python 3.12, rateslib 2.1.1

irs = IRS(
    effective=dt(2025, 11, 26), 
    termination="1y", 
    spec="sek_irs3", 
    fixed_rate=2.04, 
    leg2_fixings=2.06, 
    curves=["sek3m"],
    notional=1e9,
)
```

So now lets build two different risk models and explain their representations.

First design a simple curve with two nodes (for 3M and 1Y rates):

```
sek3m = Curve(
    nodes={dt(2025, 11, 24): 1.0, dt(2026, 2, 24): 1.0, dt(2026, 11, 30): 1.0},
    convention="act360",
    calendar="stk",
    id="sek3m",
)
```

This first risk model (1) is constructed with a 1Y IRS with fixing

```
model_1 = Solver(
    curves=[sek3m],
    instruments=[
        IRS(dt(2025, 11, 26), "3m", spec="sek_irs3", curves=["sek3m"]),
        IRS(dt(2025, 11, 26), "1y", spec="sek_irs3", curves=["sek3m"], leg2_fixings=2.06)
    ],
    s=[2.06, 2.04],
    instrument_labels=["3m", "1y"],
)

irs.delta(solver=model_1)
```

What is the interpretation here? The traded instrument exactly matches the instrument in the risk model, i.e. a 1Y IRS with fixing. The amount of risk (98k) is the genuine amount of risk that this instrument experiences per 1bp movement in the fixed rate. This representation of risk will be the same before or after the fixing, however, the trader must be aware that the volatility of the 1Y rate before the fixing and after the fixing are different. If you trade a forward swap (i.e. out of tomorrow) without a fixing, the risk will be unintuitive. I.e.

```
fwd_irs = IRS(dt(2025, 11, 27), "1y", spec="sek_irs3", fixed_rate=2.04, curves=["sek3m"], notional=1e9)
fwd_irs.delta(solver=model_1)
```

This means that a forward swap has the amount of risk similar to the 1Y with fixing but an additional amount to reflect that this forward swap hasn't yet fixed. Tomorrow, when you open you risk strip this will no longer be 125k but 100k to revert back to the above case (since tomorrow it will be a 1Y par swap pre and post fixing).

This second (2) model uses risk instruments that do not have fixings, even after 11am:

```
model_2 = Solver(
    curves=[sek3m],
    instruments=[
        IRS(dt(2025, 11, 26), "3m", spec="sek_irs3", curves=["sek3m"]),
        IRS(dt(2025, 11, 26), "1y", spec="sek_irs3", curves=["sek3m"])
    ],
    s=[2.06, 2.04],
    instrument_labels=["3m", "1y"],
)
irs.delta(solver=model_2)
```

The interpretation here is that the IRS with fixing is not the same as the instrument in the risk model, and therefore requires an adjustment in the 3M sector to mitigate risk sensitivity. Here the rate of the 'risk instrument' is not the same as the 'market par 1y' swap because it does not include the fixing, and its rate would typically need to be derived (or resampled) from the pricing curve.

The risk of the forward swap looks more appropriate (in my opinion):

```
fwd_irs.delta(solver=model_2)
```