Post
ChamperDamper
May 12, 2024
· Strategy
Strategy to decide if the temp is going up one more degree:
If you haven't read my first weather strategy post, you can find it here: https://kalshi.com/ideas/posts/a5cab4f6-0f1f-4ea9-8afc-6282e816c76c
If you trade the weather markets for any amount of time you are certain to encounter this situation:
You're watching the temperature readings closely and really need to know if the temp is going to go up one more degree and hit the next bracket.
At this point in the day, there is no room for error.
Because of rounding on the 5 minute readings, you don't know what the "real" temp is. Has it already hit between the 5 min readings? Is it about to?
Waiting for the next climate report or 6 hour high to post and expect you will be able to grab those shares faster than everyone else is unrealistic.
Here is how I approach the situation.
Let's use KMDW Chicago on May 11 as an example to see if it might have been possible to beat the market.
Here are some readings in F.
Remember, the 5 minute readings are measured in C, rounded to the nearest whole degree then converted to F. The readings at :53 every hour are rounded to the nearest tenth degree C and then converted to F.
May 11, 3:00 pm 68
May 11, 2:55 pm 71.6
May 11, 2:53 pm 69.98
May 11, 2:50 pm 69.8
May 11, 2:45 pm 69.8
May 11, 2:40 pm 69.8
May 11, 2:35 pm 69.8
May 11, 2:30 pm 66.2
Between 2:50 and 3:00 the temp went from 69.8 to 71.6 and then down to 68. That brief spike hit the top bracket in the market.
Wow, that's crazy! Or is it?
Here are the same readings in C.
May 11, 3:00 pm 20
May 11, 2:55 pm 22
May 11, 2:53 pm 21.1
May 11, 2:50 pm 21
May 11, 2:45 pm 21
May 11, 2:40 pm 21
May 11, 2:35 pm 21
May 11, 2:30 pm 19
We know the temp can fluctuate a lot in a short time. However, because of rounding it can actually be less dramatic than it seems.
Here are the last 4 readings adjusted to what I think they most likely were in C.
May 11, 3:00 pm 20.4
May 11, 2:55 pm 21.5
May 11, 2:53 pm 21.1
May 11, 2:50 pm 20.9
When the hourly came in at 69.98 F, I could infer that because the rounded temp looked steady for 20 minutes, it was on a warming trend from 68.9 towards 69.98 during that period.
And now we know the correct question to ask ourselves:
What is the probability of the temp going up another half degree in a 5 minute period during a warming trend?
To answer that, you still need to consider the conditions and time of day. However, with enough experience you can start to infer where the real temp is and how likely it is to go higher.
In this instance, I was expecting the temp to drop soon and I hesitated. However, the information was there to make a decently informed early choice.
To be clear, it's still easy to be wrong. Though I feel that by doing a little work to try and infer where the temp really is you can improve your success rate.
Strategies used inIf you haven't read my first weather strategy post, you can find it here: https://kalshi.com/ideas/posts/a5cab4f6-0f1f-4ea9-8afc-6282e816c76c
If you trade the weather markets for any amount of time you are certain to encounter this situation:
You're watching the temperature readings closely and really need to know if the temp is going to go up one more degree and hit the next bracket.
At this point in the day, there is no room for error.
Because of rounding on the 5 minute readings, you don't know what the "real" temp is. Has it already hit between the 5 min readings? Is it about to?
Waiting for the next climate report or 6 hour high to post and expect you will be able to grab those shares faster than everyone else is unrealistic.
Here is how I approach the situation.
Let's use KMDW Chicago on May 11 as an example to see if it might have been possible to beat the market.
Here are some readings in F.
Remember, the 5 minute readings are measured in C, rounded to the nearest whole degree then converted to F. The readings at :53 every hour are rounded to the nearest tenth degree C and then converted to F.
May 11, 3:00 pm 68
May 11, 2:55 pm 71.6
May 11, 2:53 pm 69.98
May 11, 2:50 pm 69.8
May 11, 2:45 pm 69.8
May 11, 2:40 pm 69.8
May 11, 2:35 pm 69.8
May 11, 2:30 pm 66.2
Between 2:50 and 3:00 the temp went from 69.8 to 71.6 and then down to 68. That brief spike hit the top bracket in the market.
Wow, that's crazy! Or is it?
Here are the same readings in C.
May 11, 3:00 pm 20
May 11, 2:55 pm 22
May 11, 2:53 pm 21.1
May 11, 2:50 pm 21
May 11, 2:45 pm 21
May 11, 2:40 pm 21
May 11, 2:35 pm 21
May 11, 2:30 pm 19
We know the temp can fluctuate a lot in a short time. However, because of rounding it can actually be less dramatic than it seems.
Here are the last 4 readings adjusted to what I think they most likely were in C.
May 11, 3:00 pm 20.4
May 11, 2:55 pm 21.5
May 11, 2:53 pm 21.1
May 11, 2:50 pm 20.9
When the hourly came in at 69.98 F, I could infer that because the rounded temp looked steady for 20 minutes, it was on a warming trend from 68.9 towards 69.98 during that period.
And now we know the correct question to ask ourselves:
What is the probability of the temp going up another half degree in a 5 minute period during a warming trend?
To answer that, you still need to consider the conditions and time of day. However, with enough experience you can start to infer where the real temp is and how likely it is to go higher.
In this instance, I was expecting the temp to drop soon and I hesitated. However, the information was there to make a decently informed early choice.
To be clear, it's still easy to be wrong. Though I feel that by doing a little work to try and infer where the temp really is you can improve your success rate.
PsecretPseudonym
May 12, 2024This is an excellent analysis and explaination. Thank you for the analysis. It sounds like you’ve found an artifact of units/rounding in how they is decide temperature contracts. Out of curiosity, have you compared weather forecasting sources for relative accuracy?
ChamperDamper
May 12, 2024@PsecretPseudonym I have not done any in depth comparison of different sources. Mostly I just stick with the NWS forecast as a starting point and then try to decide how wrong it is going to be.
GyllonDeckles
Aug 17, 2024How come everyone bets “no” on the correct answer?
ChamperDamper
Aug 17, 2024@GyllonDeckles each contract requires a yes and a no to be complete. So it is not possible for everyone to buy no.