#investing all about PE / EPS - it just can’t get simpler than this - wonderful initiative by @KirtanShahCFP to educate one and all - keep up the great work - its worth it’s weight in gold 👏👏

THE IMPORTANCE OF DEVI BEEJ / BIJA MANTRA.

Once Suutji in his discourse with King Janmajeya repeated the words of Vyasji regarding the pronounciation of the word ऐ. Vyasji had heard the importance of this word in a meeting of all important saints at Naimisharanya.

There this story was discussed.

In that gathering of such great sages, Rishi Jamdagni asked a question to all the eminent luminaries. He wanted to know which diety is the most revered and gets satisfied easily with our prayers. So Lomash rishi replied that Devi

Bhagwati was one such diety. She will be satisfied and is ready to grant you a boon even if you just say the first letter of her beej mantra. In this context he narrated the story of Satyavrat. The story is as follows.

There lived a Brahmin named Devdutt in Kosal.

He was childless. He arranged for an elaborate yagya for a intelligent son to be born. He invited all the eminent rishi's to conduct this Yagya. Rishi Suhotra, Pael, Brahaspati Yagyavalka were assigned different tasks. Rishi Gobhil was asked to sing the Saamved.

He sang the verses very well, but he had to take respite in order to breathe between two verses. This angered devdutt. He spoke his thoughts to Gobhil rishi.

The rishi got very angry as devdutt did not realise the practicalities of singing. He cursed devdutt that

Pyramiding 2.0

A thread-

What is pyramiding?

It is simply adding to your winners.

Why use pyramiding?

1. It makes your winners bigger than your losers.

• Because you are adding positions when your trades are going your way, you are increasing your gains but your losers will be small comparatively.

2. You lose less during Draw-down time ( it saves you from losing big during whipsaws )

• Many times when the market is in sideways mode, Entering a trade with full risk might not be the best idea.

• So, you test the waters with a small position first, it could be ⅓ or ¼ of your total risk.

• So if the Breakout fails or whipsaws come, then you will only lose a small part of your overall risk.

• Then when position starts to go in your favor you can add to it.

3. It helps you to concentrate on a few trades,

They say that don’t put all your eggs in one basket, but don’t put them in too many either.

• See, Everyone can see charts and find breakouts, pullbacks etc.

Important paper from Google on large batch optimization. They do impressively careful experiments measuring # iterations needed to achieve target validation error at various batch sizes. The main "surprise" is the lack of surprises. [thread]

https://t.co/7QIx5CFdfJ

The paper is a good example of lots of elements of good experimental design. They validate their metric by showing lots of variants give consistent results. They tune hyperparamters separately for each condition, check that optimum isn't at the endpoints, and measure sensitivity.

They have separate experiments where the hold fixed # iterations and # epochs, which (as they explain) measure very different things. They avoid confounds, such as batch norm's artificial dependence between batch size and regularization strength.

When the experiments are done carefully enough, the results are remarkably consistent between different datasets and architectures. Qualitatively, MNIST behaves just like ImageNet.

Importantly, they don't find any evidence for a "sharp/flat optima" effect whereby better optimization leads to worse final results. They have a good discussion of experimental artifacts/confounds in past papers where such effects were reported.