ive watched another video of yours when youve said that usually quant trading is not as glamorous and as high paying, for the most part, as online sources make it out to seem. I was really fixated on the idea of all traders as super duper millionaires right from the start, unfortunately. However, the work in credit risk actually looks quite fun to do. Im glad you did this video as all ive looked into was trading, but theres more to quant finance than just focusing on trading and this video has helped me realise that. Thanks for another great video!
Thanks for this Dimitri! Is there anymore information on the gains tables you mentioned? Interested in how you would spot those applicants who look the same but score differently. Cheers!
I actually learned it from a colleague and I've never seen it done anywhere else. I've been wanting to write a paper on it but just never got around to it. Essentially the model will score candidates the same with the same variable values however since logistic (or GBM) does not require a distribution for the residuals it is unclear if you have a missing variable problem which can cause a bias. Overfitting can be prevented by using Chi Square however missing variables are hard to detect.
I would do a quant masters like an mfe because you'll get courses more closely aligned to the industry. If you went with the applied math masters you would want to focus on statistics.
As a high schooler, what specific parts of math do you think this relies on? I know that it requires a high level understanding of calculus and probability but I'm wondering which other ones matter.
For high school you can also take linear algebra and basic statistics. Once you get to college and grad school you'll want to cover ode, pde, real analysis, and sde for a well rounded quant background. I find statistics as a whole the main focus of quant finance though.
never done risk modeling but hearing you talk about credit risk sounds pretty interesting to model. will pick up Siddiqi's book and maybe work on a credit risk modeling project independently. too bad i can't afford SAS as an individual though lol
Yes and no. Models do provide a more accurate and fair assessment of lending however there are some trade-offs. Models are expensive because you need people to develop and validate them. The quants building models start at $70-100k and the average person will be making around $140k. Small banks usually don't do enough business to make this worth while. You also have to convince people that model work. Many people not used to using models are skeptical which is understandable. You also get what you pay for. Cheap developers will build poor quality models and you could have larger losses from this. The other constraint is data. You need a lot of data to build good models. Firms strategy of issuing loans also effects the data that they should be using. If you do a few thousand loans a year in either retail or commercial, you won't have enough data to build reliable models. Models also have to be rebuilt over time as strategy changes and the economy changes. There isn't really a good way to build a model that works for everyone or for a long time.
@@DimitriBianco I wish there was a way to implement a little more analysis in the traditional roles that I’ve had, but even if there was one, I think it will be a long time before any change. I’ve also ran into the “convincing” of people that some models are useful. Thanks for your answer!
ive watched another video of yours when youve said that usually quant trading is not as glamorous and as high paying, for the most part, as online sources make it out to seem. I was really fixated on the idea of all traders as super duper millionaires right from the start, unfortunately. However, the work in credit risk actually looks quite fun to do. Im glad you did this video as all ive looked into was trading, but theres more to quant finance than just focusing on trading and this video has helped me realise that. Thanks for another great video!
Thanks for this Dimitri! Is there anymore information on the gains tables you mentioned? Interested in how you would spot those applicants who look the same but score differently. Cheers!
I actually learned it from a colleague and I've never seen it done anywhere else. I've been wanting to write a paper on it but just never got around to it. Essentially the model will score candidates the same with the same variable values however since logistic (or GBM) does not require a distribution for the residuals it is unclear if you have a missing variable problem which can cause a bias. Overfitting can be prevented by using Chi Square however missing variables are hard to detect.
should i choose an applied math masters or an mfe to have better chances to become quant?
I would do a quant masters like an mfe because you'll get courses more closely aligned to the industry. If you went with the applied math masters you would want to focus on statistics.
As a high schooler, what specific parts of math do you think this relies on? I know that it requires a high level understanding of calculus and probability but I'm wondering which other ones matter.
For high school you can also take linear algebra and basic statistics. Once you get to college and grad school you'll want to cover ode, pde, real analysis, and sde for a well rounded quant background. I find statistics as a whole the main focus of quant finance though.
never done risk modeling but hearing you talk about credit risk sounds pretty interesting to model. will pick up Siddiqi's book and maybe work on a credit risk modeling project independently. too bad i can't afford SAS as an individual though lol
They have a student version of SAS that is free. I've been thinking about downloading it and doing some SAS videos.
@@DimitriBianco that would be awesome. i really learned a lot from your R videos
@@DimitriBianco would also be interested in some SAS videos!
Do traditional finance credit underwriting departments eventually evolve into having more sophisticated quantitative analyses?
Yes and no. Models do provide a more accurate and fair assessment of lending however there are some trade-offs. Models are expensive because you need people to develop and validate them. The quants building models start at $70-100k and the average person will be making around $140k. Small banks usually don't do enough business to make this worth while. You also have to convince people that model work. Many people not used to using models are skeptical which is understandable. You also get what you pay for. Cheap developers will build poor quality models and you could have larger losses from this.
The other constraint is data. You need a lot of data to build good models. Firms strategy of issuing loans also effects the data that they should be using. If you do a few thousand loans a year in either retail or commercial, you won't have enough data to build reliable models. Models also have to be rebuilt over time as strategy changes and the economy changes. There isn't really a good way to build a model that works for everyone or for a long time.
@@DimitriBianco I wish there was a way to implement a little more analysis in the traditional roles that I’ve had, but even if there was one, I think it will be a long time before any change. I’ve also ran into the “convincing” of people that some models are useful.
Thanks for your answer!