Are you aware that your model may be biased?

16 October 2017

Analytical models and their inherent bias

Energy companies need to value and manage risks in their portfolios. To achieve this, advanced analytical models are key. Is your company using an in-house developed model? If the answer is yes, are you aware of the risk that your model may be biased? We propose to use industry standard KYOS analytical models – either for your valuations, or use them to benchmark your in-house developed models.

Are your valuations biased?

The portfolio of energy companies contains many assets and contracts with flexibility. In order to properly value and risk manage in these portfolios, sophisticated analytical models are being used. These models are typically based on advanced mathematical theories and often include Monte Carlo simulations. Some companies see the development, implementation and maintenance of such models as a core competence and as such decide to do these activities in-house. Every day, many quant developers across the industry work hard at these complex models, calibrating the model parameters to market data. Models that often get so complicated that only the quant developers themselves fully understand them and, more importantly, understand their limitations.

Risk of model bias

Big and complex portfolios are being managed and hedged based on these internally developed models. Risk capital is allocated, risk limits set and hedge strategies are defined, all based on the same set of models. Moreover, based on the same model results, management decides and shareholders are informed on the P&L and risk position of their company. Would you really want that the decisions you take rely only on in-house developed models? Do you actually know how different they are to the models in other companies? Are you aware that models for different applications often use the same basis, not rarely developed by the same developers and therefore all exhibit similar behavior? What would be the impact on your company if the in-house models exhibit a structural bias? What if these models always underestimate the risk in certain price scenarios?

Bench-marking is best practice

Bench-marking internally created models with one or more externally developed models is the norm in the financial sector: already before the economic crisis of 2009, but this behavior got even more in use after 2009. This bench-marking can be done in various ways. Typically, banks run external valuations on a regular basis for a portfolio with a set of key contract types. In order to stress-test the models, these benchmarks can be done for a range of historical market prices to mimic many different market scenarios. In this way you can discover weaknesses of the model and obtain a better understanding of the validity of the risk numbers. If necessary, companies adjust their own models based on these external benchmarks.

Given the current complexity of the portfolios of energy trading companies, bench-marking models with other models should become best practice in the energy sector as well.

How can KYOS support

KYOS has developed best in class analytical models for most of the commonly used energy contract types. KYOS analytical models are widely accepted in the market and in use by energy companies and traders across Europe. The models are integrated in the KYOS Analytical Platform for an easy user experience or can be integrated into existing ETRM systems.

KYOS can support your company in three ways to avoid bias of your valuations and risk reports:

  • Use KYOS analytical models as the standard models in your company. This ensures that your company has valuations that are in-line with the industry. Separately, companies can benchmark the KYOS models with in-house developed tools.
  • Benchmark your existing models with KYOS analytical models. Use the KYOS models in parallel to your existing models to identify valuation differences in an early stage, or run them on a regular basis on a test portfolio.
  • KYOS can perform a bench-marking study of your existing models. KYOS will test your existing models under various market conditions, and compare these results to KYOS’ models. You will get tailored advice on where and how you models might be biased.

Avoid the bias trap and start using KYOS’ full range of analytical models!

For more information on how KYOS can support your company in validating models, fill in the contact form below or contact Ewout Eijkelenboom directly at Ewout@kyos.com

Contact KYOS

KYOS
Nieuwe Gracht 49
2011 ND Haarlem
The Netherlands

E-mail: info@kyos.com
Tel: +31 (0)23 551 02 21

Follow us

Contact us via the form below. We will reply within 48 hours.

Request a demo via the form below. We will reply within 48 hours.

  • This field is for validation purposes and should be left unchanged.

Book a training course via the form below.

  • (If available)
  • This field is for validation purposes and should be left unchanged.

Request a training course via the form below. We will reply within 48 hours.

  • This field is for validation purposes and should be left unchanged.

Plan a meeting via the form below. We will reply within 48 hours.

  • This field is for validation purposes and should be left unchanged.