BDI & SCFI prediction (‘22/18W)

In this article, I'm going to compare the superiority of the models by periodically predicting/verifying the freight rates using ARIMA and logistic regression models.

---

1.     Last Week’s forecasts

1-1.        BDI, Actual Vs. Predicted

Date	Last week Forecast	Actual BDI	MAPE(%)	Accuracy(%)
‘22/17W(4/25~29)	2,246.158	2,398.40	6.3%	93.7%

 

1-2.        SCFI, Actual Vs. Predicted

Date	Last week Forecast	Actual SCFI	MAPE(%)	Accuracy(%)
‘22/17W(4/25~29)	4,187.446	4,177.30	0.2%	99.8%

 The prediction accuracy of BDI and SCFI was 93.7% and 99.8%, respectively. The predicted values (2,246.158) and actual values (2,398.40) of BDI, and predicted values (4,187.446) and actual values (4,177.30) of SCFI all achieved Trend and Accuracy. The average BDI MAPE value for the last 5 predictions is 3.3% and the prediction rate is 96.7%, and the SCFI average MAPE value is 0.9% and the prediction rate is 99.1%.

---

2.     BDI forecast

2-1. ARIMA, Time series analysis

 For the BDI index forecast, I used weekly data (430 pieces) from January 3, 2014 to April 29, 2022 to predict the freight for three weeks in the future.

Date	Forecast	Trend(WoW)
‘22/18W(5/2~6)	2,511.710	↑
‘22/19W(5/9~13)	2,535.977	↗
‘22/20W(5/16~20)	2,523.284	↘

 As shown in the table and figure above, BDI (as of 4/29: 2,398.40) increased by 9% compared to last week (as of 4/22: 2,200.65), showing a clear upward trend. In particular, considering that the current Cape market is very low compared to Panamax, and the 3Q peak season, the BDI market is expected to rise gradually.
In the ARIMA model, after rising up to the 19th week, it was found that the freight rate remained somewhat flat at the 20th week.

Ref) BDI & SCFI prediction (‘22/17W) (tistory.com)

BDI & SCFI prediction (‘22/17W)

 

2-2. Logistic regression analysis

 Logistic regression analysis was performed with the same data predicted by the ARIMA model previously.

 The predicted trend for next week was 0.654, higher than the baseline 0.5, and higher than the previous week's forecast trend of 0.566. In other words, the direction of the dry bulk market next week is expected to rise to a very high level.

---

3.     SCFI Outlook

3-1. ARIMA, Time series analysis

 For the SCFI index forecast, I used weekly data (628 pieces) from October 16, 2009 to April 29, 2022 to predict the future freight for three weeks.

Date	Forecast	Trend(WoW)
‘22/18W(5/2~6)	4,164.260	↘
‘22/19W(5/9~13)	4,156.429	→
‘22/20W(5/16~20)	4,147.861	→

 

 As shown in the table and figure above, the SCFI (as of 4/29: 4,177.30) is expected to show a continuous decline following the previous week's forecast. However, considering the trend of the decline continuing to decrease, and from a statistical point of view, it appears to be nearing the bottom on the chart.

 One thing to be aware of is that when it approaches the bottom like this, if there is an upturn signal, the freight rate may rise to a higher level than expected, so you have to keep an eye on it.

 

3-2. Logistic regression analysis

 The results of the logistic regression analysis model of SCFI are also the same as the ARIMA values. With a predicted value of 0.385, the next week's SCFI freight direction is predicted to be down.

 

Thanks.