An argument for why the Gambler's Fallacy, the Recency Bias, and Conditional Heteroskedasticity could set up an attractive asymmetric return profile ahead of Wednesday's US inflation (#CPI) data ...
🧵 1/n
🧵 1/n
Headline CPI is forecasted to come in at 6.5% YoY, compared to 7.1% the month before. CPI has clearly been trending downward since June 2022, which is positive for the #economy and risky assets as it puts a ceiling on how high the #FederalReserve has to raise rates.
2/n
2/n
The Gambler's Fallacy & Recency Bias:
CPI swaps indicate that CPI should be approx. 2.5% (!) this June. This seems to be an example of the "Gambler's Fallacy"; a misunderstanding of #probabilities in which people wrongly project reversal towards a long-term mean.
3/n
CPI swaps indicate that CPI should be approx. 2.5% (!) this June. This seems to be an example of the "Gambler's Fallacy"; a misunderstanding of #probabilities in which people wrongly project reversal towards a long-term mean.
3/n
Enter "Recency Bias": It seems the market is extrapolating the recent positive CPI data.
The #market might be right but forecasting #inflation is no easy task. One year ago economists forecasted CPI would be at 3% (!) today. Consensus of 2.5% by June might be doable but ...
4/n
The #market might be right but forecasting #inflation is no easy task. One year ago economists forecasted CPI would be at 3% (!) today. Consensus of 2.5% by June might be doable but ...
4/n
... it definitely seems like it's at the optimistic end of possible CPI outcomes!⚠️
This could leave the market vulnerable to a negative surprise in CPI, especially when we take into account that previously, on the day CPI has been released, the market seems to ...
5/n
This could leave the market vulnerable to a negative surprise in CPI, especially when we take into account that previously, on the day CPI has been released, the market seems to ...
5/n
... have been reacting to how CPI printed compared to the forecast and NOT to the prior month's/trend's data.
Conditional Heteroskedasticity:
CPI exhibits conditional heteroskedasticity, i.e., non-constant #variance ...
6/n
Conditional Heteroskedasticity:
CPI exhibits conditional heteroskedasticity, i.e., non-constant #variance ...
6/n
... basically, that means there is a random component to forecasting CPI, especially when CPI is at elevated levels as the #variability is greater here.
The forecasted CPI value is often the center of a distribution of possibilities, but as CPI is a random variable ...
7/n
The forecasted CPI value is often the center of a distribution of possibilities, but as CPI is a random variable ...
7/n
... it can take on any value that falls within the range of its #distribution.
E.g., if CPI comes in 0.2 greater than expected, it will still fall within a 95% confidence range, meaning a statistical test would show it is statistically NOT different from the expected value.
8/n
E.g., if CPI comes in 0.2 greater than expected, it will still fall within a 95% confidence range, meaning a statistical test would show it is statistically NOT different from the expected value.
8/n
An asymmetric return profile:
As the market might be overly optimistic about the pace of slowing inflation, AND as the market reacts to how CPI data lines up compared to the #forecast, which is influenced by a random component, I believe we’re facing an event with ...
9/n
As the market might be overly optimistic about the pace of slowing inflation, AND as the market reacts to how CPI data lines up compared to the #forecast, which is influenced by a random component, I believe we’re facing an event with ...
9/n
... an asymmetric return profile that could be attractive to #trade!
If CPI comes in worse than the forecast, I think the market reaction will be more severe (a larger #selloff) than if CPI comes in better than the forecast (a small #rally, if any). I.e., asymmetric! 🤩
10/n
If CPI comes in worse than the forecast, I think the market reaction will be more severe (a larger #selloff) than if CPI comes in better than the forecast (a small #rally, if any). I.e., asymmetric! 🤩
10/n
Therefore, statistically, it should be attractive to reduce beta/risk ahead of the CPI data if you don't have a clear view on how CPI will print compared to the forecast - which to a large degree is random!
Friday’s rally in $SPY and $UST bonds on the back of weaker than
11/n
Friday’s rally in $SPY and $UST bonds on the back of weaker than
11/n
... expected wage growth (which I think could be short-lived) further supports my thesis the market might be ripe for a move lower if a catalyst like a negative CPI surprise shows up.
I'll be by buying puts on $SPY, and I'll be switching some $TLT exposure to $TIPS.
12/12
I'll be by buying puts on $SPY, and I'll be switching some $TLT exposure to $TIPS.
12/12
• • •
Missing some Tweet in this thread? You can try to
force a refresh
