Using AI to help write SDQL

[A B]

Folks,

I was curious to see how well AI (Chat gpt) could handle, me inputing a plain English trend. and asking it to convert that to proper sdql. It did a pretty decent job.

Here is what I typed exactly into perplexity.ai  :

" Using the directions on how to write sdql found here: https://killersports.com/query?filter=MLB and here https://sdql.com/ and here https://groups.google.com/g/SportsDataBase , could you please write this MLB system from plain english to sdql: Play Against - Play On - Any team (PHILADELPHIA) - very good offensive team (>=5.0 runs/game) against an average starter (ERA= 4.20 to 5.20)-NL, with a slugging percentage of .390 or worse over their last 20 games "

-----------------------------

It replied:

Answer

To translate the given MLB system from plain English to SDQL, we need to break down the conditions and parameters specified:

1. Play Against: This indicates the team we are betting against.
2. Play On: This indicates the team we are betting on.
3. Any team (PHILADELPHIA): The team we are focusing on is Philadelphia.
4. Very good offensive team (>=5.0 runs/game): The opposing team must have an average of 5 or more runs per game.
5. Against an average starter (ERA= 4.20 to 5.20): The opposing team's starting pitcher must have an ERA between 4.20 and 5.20.
6. NL: The game must be in the National League.
7. With a slugging percentage of .390 or worse over their last 20 games: The opposing team must have a slugging percentage of .390 or lower in their last 20 games.

Here is the SDQL query that captures all these conditions:

sdql

team != 'PHI' and o:team = 'PHI' and o:runs >= 5 and starter:ERA >= 4.20 and starter:ERA <= 5.20 and league = 'NL' and o:SLG(last 20 games) <= .390

Explanation of the SDQL query:

• team != 'PHI': We are playing against teams that are not Philadelphia.
• o:team = 'PHI': The opponent is Philadelphia.
• o:runs >= 5: The opponent (Philadelphia) has an average of 5 or more runs per game.
• starter:ERA >= 4.20 and starter:ERA <= 5.20: The starting pitcher's ERA is between 4.20 and 5.20.
• league = 'NL': The game is in the National League.
• o:SLG(last 20 games) <= .390: The opponent's slugging percentage over the last 20 games is .390 or worse.

This query should provide the desired data based on the specified conditions.

[John Ryan]

That is interesting, no doubt, but to my knowledge there is no shortcut for slugging percentage (SLG). It would be great if we did have someone create a shortcut for SLG and OPS and OPS+   along with FIP too.   Anyway, if you copy and paste the SDQL it will not work and wondering if the AI can be trained first to know that SLG is a formula and starter ERA is a formula. etc. 

--
You received this message because you are subscribed to the Google Groups "SportsDataBase" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sportsdatabas...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sportsdatabase/f796a7ed-6381-40a5-920a-b705106a5782n%40googlegroups.com.

[Nic Feliccia]

TO my knowledge chat GPT doesn't use Retrieval Augmented Generation ; meaning it doesn't read instructions.  But apparently Perplexity does. 

I used a model tuning approach with chat GPT 3.5. 

---

From: sportsd...@googlegroups.com <sportsd...@googlegroups.com> on behalf of A B <ther...@gmail.com>
Sent: Monday, June 17, 2024 4:07 PM
To: SportsDataBase <sportsd...@googlegroups.com>
Subject: Using AI to help write SDQL

 

--